evaluation of adsorptive and biological mode dbp removal ... · evaluation of adsorptive and...
TRANSCRIPT
University of Colorado, BoulderCU Scholar
Civil Engineering Graduate Theses & Dissertations Civil, Environmental, and Architectural Engineering
Spring 1-1-2017
Evaluation of Adsorptive and Biological Mode DbpRemoval in Activated Carbon FiltersNathan YangUniversity of Colorado at Boulder, [email protected]
Follow this and additional works at: https://scholar.colorado.edu/cven_gradetds
Part of the Civil Engineering Commons, and the Environmental Engineering Commons
This Thesis is brought to you for free and open access by Civil, Environmental, and Architectural Engineering at CU Scholar. It has been accepted forinclusion in Civil Engineering Graduate Theses & Dissertations by an authorized administrator of CU Scholar. For more information, please [email protected].
Recommended CitationYang, Nathan, "Evaluation of Adsorptive and Biological Mode Dbp Removal in Activated Carbon Filters" (2017). Civil EngineeringGraduate Theses & Dissertations. 174.https://scholar.colorado.edu/cven_gradetds/174
Evaluation of Adsorptive and Biological Mode DBP Removal in Activated Carbon Filters
by
Nathan Yang B.S., University of California at Davis, 2014
A thesis submitted to the Faculty of the Graduate School of the
University of Colorado at Boulder in partial fulfillment of the requirement for the degree of
Master of Science Department of Civil, Environmental and Architectural Engineering
2016
This thesis entitled:
Evaluation of Adsorptive and Biological Mode DBP Removal in Activated Carbon Filters
written by Nathan T. Yang has been approved for the
Department of Civil, Environmental, and Architectural Engineering
R. Scott Summers (chair)
Christopher Corwin
Chad Seidel
August 19, 2016
The final copy of this thesis has been examined by the signatories, and we find that both the content and the form meet acceptable presentation standards
of scholarly work in the above mentioned discipline.
iii
Abstract Yang, Nathan T. (M.S., Environmental Engineering)
Evaluation of Adsorptive and Biological DBP Removal in Activated Carbon Filters
Thesis directed by R. Scott Summers, Professor, Department of Civil, Environmental, and Architectural Engineering, University of Colorado at Boulder Small drinking water systems face unique compliance challenges with regards to
many water quality parameters, including disinfection-by-product (DBP) levels in the
distribution system. Filtration with granular activated carbon (GAC) can be an effective
technology for the removal of total organic carbon (TOC) and DBPs.
The objectives of this thesis were to develop and evaluate the use of GAC in the
distribution system to meet DBP regulations under both adsorptive and biological modes.
It was hypothesized that a post-treatment reactor strategically located in the distribution
system will offer small systems a cost-effective alternative to controlling total
trihalomethanes (TTHMs), sum of five haloacetic acids (HAA5s) and other unregulated
DBPs. A total of six adsorptive rapid small scale column tests (RSSCTs) and three pilot
scale biofilters were operated to investigate the effects of GAC type, source water
quality, temperature and empty bed contact time (EBCT) on the adsorption and
biodegradation of TOC and DBPs in treated drinking water.
Experimental results show that adsorption with bituminous GAC is an effective
treatment strategy for the removal of TOC and TTHMs through at least 6,000 bed
volumes (42 days at 10min EBCT) and often longer depending on influent conditions.
Pore surface diffusion model (PSDM) analysis indicated that the presence of both natural
organic matter (NOM) and co-solutes are important to consider when analyzing THM
breakthrough, with THM adsorbability being the most important factor in determining
iv
breakthrough order (TCM à DCBM à DBCM à TBM) and influent concentration
determining localized breakthrough. Experimental HAA adsorption results were
nonsystematic.
In biofiltration pilot runs, DCAA and TCAA made up >85% of HAA5.
Experimental DCAA removal between 83%-97% was reported at all EBCTS (5, 10 and
20min) for the duration of the pilot runs. TCAA removal ranged between 50%-78% at 5
minute EBCT, 80%-96% at 10 minute EBCT and 93%-98% at 20 minute EBCT. No
THM biodegradation was observed. HAA reduction and reformation results indicated
that biofiltration is an effective treatment for the reduction in HAA5 both immediately
after biofiltration as well as at the end of the distribution system, across many ranges of
chlorinated influent bromide and TOC conditions.
v
Acknowledgments
Graduate school was never in my plans. That was until a co-op position at the
Central Contra Costa Sanitary District sparked my interest in research. The group of
Samantha Engelage, Michael Cunningham and Michael Falk were my first lab mates and
as we shared a wastewater trailer for many months, I became an environmental engineer.
The American Water Works Association Carollo Engineers Scholarship,
California Water Environment Association Kirt Brooks Memorial Water Environment
Scholarship, American Public Works Association Scholarship and the Environmental
Engineers of the Future Program generously provided the financial means for me to take
the leap and attend graduate school.
I would like to thank the faculty at CU Boulder as well as my past and present lab
mates and colleagues that never failed to lend a hand in the lab or a word of
encouragement. The reason I came to CU Boulder: my advisor, Scott Summers, who put
me in a position to succeed. Chris Corwin, the most caring educator I have ever
encountered, and a man who I have utmost respect for. Dorothy Noble, Leigh Terry, Kyle
Shimabuku, Anthony Kennedy and Eli Townsend, who taught me the ways of the lab.
Garrett McKay and Mandi Hohner for averting disaster on multiple occasions.
Chad Seidel for being a part of my committee. Eric Dickenson at Southern Nevada Water
Authority for providing bioGAC media. My office mates Riley Mulhern and Paige
Pruisner, as well as my roommates Anna McKenna and Scott Singer for all the
intangibles.
I would never have chosen engineering as a major if it weren’t for my best friend
Glen Lischeske, who has been by my side every step of the way. A true friend who keeps
vi
me grounded and never fails to point out the “bright” side of a situation. My brother Tim,
who’s hard work is an inspiration to me, and all around him. I am immeasurably grateful
for my parents, the most loving people I know, and Jeannie Darby - my guardian angel.
This research was funded by the EPA National Center for Innovation in Small
Drinking Water Systems as part of the Design of Risk-reducing, Innovative-
Implementable Small-System Knowledge (DeRISK) Center project (EPA-G2013-STAR-
G1).
vii
Table of Contents
LIST OF TABLES IX LIST OF FIGURES X CHAPTER 1 1 INTRODUCTION 1 1.1 MOTIVATION 1 1.2 RESEARCH OBJECTIVES 3 1.3 THESIS ORGANIZATION 4
CHAPTER 2 5 BACKGROUND 5 2.1. DISINFECTION BY PRODUCT FORMATION AND CONTROL 5 2.2. ADSORPTION BY GRANULAR ACTIVATED CARBON 6 2.2.1. TOC Adsorption 9 2.2.2. THM Adsorption 10 2.2.3. HAA Adsorption 16
2.3. BIOLOGICAL ACTIVITY IN GAC FILTERS 16 2.3.1 TOC Biodegradation 17 2.3.2 THM Biodegradation 17 2.3.3 HAA Biodegradation 18
CHAPTER 3 22 MATERIALS AND METHODS 22 3.1 MATERIALS 22 3.1.1 Activated Carbon Specifications 22 3.1.2 Source Waters 23 3.1.3 Chemicals 23
3.2 METHODS 24 3.2.1 Analytical Lab Methods 24 3.2.2 Rapid Small Scale Column Tests (after Kempisty, 2014) 25 3.2.3 Fixed-‐Bed Adsorption Modeling (after Kempisty, 2014) 30 3.2.4 Biofilter Pilot Column Tests (after Zearley, 2012) 32
CHAPTER 4 34 RESULTS AND DISCUSSION 34 4.1 EFFECT OF GAC TYPE (RSSCT #1) 34 4.1.1 TOC Adsorption 35 4.1.2 DBP Removal 38
4.2 EFFECT OF SOURCE WATER QUALITY (RSSCT #2) 44 4.2.1 TOC Adsorption 47 4.2.2 DBP Removal 51 4.2.3 Effect of Influent TOC on TTHM Breakthrough 57 4.2.4 Effect of EBCT on THM Breakthrough 58 4.2.5 Effect of Influent Concentration on TTHM Breakthrough 60 4.2.6 Relative Effects of NOM and Co-‐solutes on THM Breakthrough 63
4.3 EFFECT OF TEMPERATURE, INFLUENT BROMIDE AND INFLUENT TOC ON BIODEGRADATION OF DBPS (PILOT RUNS #1 AND #2) 73 4.3.1 Biomass Distribution Throughout Pilot Operation 75
viii
4.3.2 TOC Removal 77 4.3.3 Pseudo First Order Rate Equation 78 4.3.4 HAA Biodegradation 80 4.3.5 Effect of Temperature of HAA Biodegradation 86 4.3.6 THM Biodegradation and Reformation 88 4.3.7 HAA Reformation and Treatment Effectiveness 88
4.4 SUMMARY OF RESULTS 93 4.4.1 Adsorption 93 4.4.2 Biodegradation 95
CHAPTER 5 97 SUMMARY AND RECOMMENDATIONS 97
WORKS CITED 102 APPENDIX A – GAC MANUFACTURER SPECIFICATIONS 106 APPENDIX B – TOC ADSORPTION 115 APPENDIX C – THM ADSORPTION 122 APPENDIX D – HAA ADSORPTION 126 APPENDIX E – TOC BIODEGRADATION 129 APPENDIX F – THM BIODEGRADATION AND REFORMATION 131 APPENDIX G – HAA BIODEGRADATION AND REFORMATION 133 APPENDIX H – ATP BIOMASS MEASUREMENTS AND METHOD 140
ix
List of Tables Table 1-1: Stage 2 DBPR MCLs and MCLGs .................................................................... 2 Table 2-1: Comparison of Physical and Chemical Adsorption (adapted from Crittenden et
al., 2012) ..................................................................................................................... 7 Table 2-2: Trihalomethane Adsorption Affinity Indicators for Bituminous based GAC
(Speth & Miltner, 1990; World Health Organization, 2004) .................................... 10 Table 2-3: THM Breakthrough Literature Review ........................................................... 12 Table 2-4: HAA Biodegradation Literature Review ......................................................... 19 Table 3-1: Granular Activated Carbon Manufacturer Specifications ............................... 22 Table 3-2: Source Water Quality ...................................................................................... 23 Table 3-3: Analytical Methods ......................................................................................... 24 Table 4-1: RSSCT #1 Influent Characteristics ................................................................. 35 Table 4-2: Trihalomethane Adsorption Affinity Indicators for Bituminous based GAC
(Speth & Miltner, 1990; World Health Organization, 2004) .................................... 38 Table 4-3: HAA Adsorption Affinity Indicators (Speth & Miltner, 1990; World Health
Organization, 2004) .................................................................................................. 39 Table 4-4: Bed Volumes to 50% Breakthrough (BV50) .................................................... 42 Table 4-5: Average influent Water Characterization ........................................................ 44 Table 4-6: Bed Volumes to 50% Breakthrough (BV50) of TOC at influent TOC
concentration of 2.1-2.3 mg/L .................................................................................. 49 Table 4-7: 50% Breakthrough Values (Bed Volumes x 103) ............................................ 56 Table 4-8: BT 10min EBCT Model and Experimental Breakthrough .............................. 70 Table 4-9: BTCl2 10min EBCT Model and Experimental Breakthrough ......................... 71 Table 4-10: BTBr 10min EBCT Model and Experimental Breakthrough ........................ 71 Table 4-11: Reverse BTBr Concentration Model Breakthrough ...................................... 71 Table 4-12: Same Concentration Model Breakthrough .................................................... 72 Table 4-13: Influent Conditions ........................................................................................ 74 Table 4-14: TOC Removal across 20 minute EBCT for all six influent conditions ......... 78 Table 4-15: Influent HAA Concentrations ....................................................................... 80 Table 4-16: Extrapolated Contaminant Utilization Rate Constants .................................. 83
x
List of Figures Figure 1-1: Remote GAC treatment schematic ................................................................... 3 Figure 2-1: GAC Contactor Schematic: Idealized Adsorption Zone and Resulting
Breakthrough (Noto, 2016) ......................................................................................... 8 Figure 2-2: Representative DOC breakthrough for activated carbon columns (Summers et
al., 2010) ..................................................................................................................... 9 Figure 2-3: Influent TOC vs TTHM BV50 ........................................................................ 15 Figure 2-4: Effect of Temperature and EBCT on HAA Biodegradation (data from
references in Table 2-4) ............................................................................................ 20 Figure 3-1: Base RSSCT Set Up (after Kempisty 2014) .................................................. 26 Figure 3-2: Biofilter Setup ................................................................................................ 33 Figure 4-1: TOC Breakthrough at 5min EBCT for three different GAC types - (Inf. TOC
= 1.3 mg/L) ............................................................................................................... 36 Figure 4-2: TOC Breakthrough at 10min EBCT for three different GAC types - (Inf.
TOC= 1.3 mg/L) ....................................................................................................... 37 Figure 4-3: TTHM, Speciated THM and TOC Breakthrough - Bituminous 10min EBCT
................................................................................................................................... 40 Figure 4-4: TTHM, Speciated THM and TOC Breakthrough - Lignite 10min EBCT ..... 41 Figure 4-5: TTHM, Speciated THM and TOC Breakthrough - Coconut 10min EBCT ... 42 Figure 4-6: TTHM Breakthrough at 10min EBCT – Carbon Type (Inf. TTHM = 28.5
µg/L) ......................................................................................................................... 43 Figure 4-7: Influent TTHM Concentration Gradient - Influent Chlorine and Bromide) .. 45 Figure 4-8: Model and Experimental Breakthrough of TCM at 10min EBCT for the BT
and BTCl2 waters ...................................................................................................... 46 Figure 4-9: Model and Experimental Normalized Breakthrough at 10min EBCT ........... 47 Figure 4-10: TOC Breakthrough at 5min EBCT for three influent conditions – BT, BTCl2
and BTBr ................................................................................................................... 48 Figure 4-11: TOC Breakthrough at 10min EBCT for three influent conditions – BT, BT
with added Chlorine and BT with added Bromide ................................................... 49 Figure 4-12: TOC Breakthrough at 5, 10 and 20min EBCT for the BTBr water ............. 50 Figure 4-13: TTHM, Speciated THM and TOC Breakthrough - BT 10min EBCT ......... 52 Figure 4-14: TTHM, Speciated THM and TOC Breakthrough - BTCl2 10min EBCT .... 53 Figure 4-15: TTHM, Speciated THM and TOC Breakthrough - BTBr 5min EBCT ....... 54 Figure 4-16: TTHM, Speciated THM and TOC Breakthrough - BTBr 10min EBCT ..... 54 Figure 4-17: TTHM, Speciated THM and TOC Breakthrough - BTBr 20min ................ 55 Figure 4-18: TTHM Breakthrough - Influent Chlorine and Bromide ............................... 55 Figure 4-19: Effect of Influent TOC on THM Breakthrough at 10min EBCT – Boulder
Tap Water from RSSCT #1 and RSSCT #2 .............................................................. 58 Figure 4-20: Experimental Effect of EBCT on TCM Breakthrough - BTBr water .......... 59 Figure 4-21: Experimental Effect of EBCT on DCBM Breakthrough – BTBr water ...... 59 Figure 4-22: Single Solute Modeled EBCT Effect on TCM Breakthrough in Organic Free
Water ......................................................................................................................... 60 Figure 4-23: Modeled Single-solute TCM Breakthrough at 10min EBCT at different
influent concentrations .............................................................................................. 61 Figure 4-24: Experimental TCM Breakthrough at 10min EBCT ..................................... 62
xi
Figure 4-25: Modeled Co-Solute TCM Breakthrough at 10min EBCT at three influent concentrations ........................................................................................................... 63
Figure 4-26: Modeled Single-Solute THM Relative Breakthrough ................................. 64 Figure 4-27: Single-solute, Co-solute Breakthrough and NOM-Solute for TCM and
DBCM- PSDM Model .............................................................................................. 65 Figure 4-28: Single-solute, Co-solute Breakthrough and NOM-Solute for DCBM and
TBM- PSDM Model ................................................................................................. 66 Figure 4-29: BT 10min ECBT Model and Experimental TTHM Breakthrough (Inf TTHM
= 58.5 µg/L) .............................................................................................................. 68 Figure 4-30: BTCl2 10min EBCT Model and Experimental Breakthrough (Inf TTHM =
85.9 µg/L) ................................................................................................................. 69 Figure 4-31: BTBr 10min EBCT Model and Experimental Breakthrough (Inf TTHM =
65.3 µg/L) ................................................................................................................. 70 Figure 4-32: Experimental Setup ...................................................................................... 74 Figure 4-33: Biomass Distribution in Chlorinated Influent Biofilter ............................... 76 Figure 4-34: DCAA Removal as a function of EBCT for all six influent conditions at 21°
C ................................................................................................................................ 81 Figure 4-35: DCAA Removal as a function of total biomass activity for all six influent
conditions at 10 and 21 °C ........................................................................................ 82 Figure 4-36: TCAA Removal as a function of EBCT for all six influent conditions ....... 84 Figure 4-37: TCAA Removal as a function of total biomass activity for all six influent
conditions at 10 and 21 °C ........................................................................................ 85 Figure 4-38: Temperature Effects on DCAA Removal .................................................... 87 Figure 4-39: Temperature Effects on TCAA Removal ..................................................... 87 Figure 4-40: HAA Reformation - 0 microgram/L Br Influent .......................................... 89 Figure 4-41: HAA Reformation - 50 microgram/L Br Influent ........................................ 90 Figure 4-42: HAA Reformation - 100 microgram/L Br Influent ...................................... 90 Figure 4-43: HAA Reformation - 1mg/L TOC Influent ................................................... 91 Figure 4-44: HAA Reformation - 2 mg/L TOC Influen ................................................... 91 Figure 4-45: HAA Reformation - 3.5 mg/L TOC Influent ............................................... 92 Figure 4-46: HAA5 Reformation - Effect of Influent Bromide and TOC ........................ 92
1
Chapter 1 Introduction
1.1 Motivation
As communities grow and the drinking water networks get bigger, the
amount of time that water spends in the distribution network before it reaches customers
can increase, creating challenges to maintaining water quality. The most prevalent
chlorinated disinfection-by-products (DBPs) in drinking water are the four species of
trihalomethanes (chloroform (TCM), dichlorobromomethane (DCBM),
dibromochloromethane (DBCM), and bromoform (TBM)) and nine species of haloacetic
acids (HAAs). Total trihalomethanes (TTHM), the sum of all four trihalomethanes
(THMs), as well as the sum of five haloacetic acids (HAA5), i.e., monochloroacetic acid
(MCAA), monobromoacetic acid (MBAA), dichloroacetic acid (DCAA), dibromoacetic
acid (DBAA), and trichloroacetic acid (TCAA), are regulated in the United States
(USEPA, 2015). Promulgated January 5, 2006, the stage 2 disinfectants and disinfection
by products rule (DBPR) strengthened regulation of TTHM and HAA5 from the previous
stage 1 DBPR. Compliance monitoring for TTHM and HAA5 was changed from a
distribution system running average to a locational running average, meaning that the
maximum contaminant levels (MCLs) shown in Table 1-1 will be calculated for each
monitoring location in the distribution system as opposed to an average of all distribution
system monitoring points.
2
Table 1-1: Stage 2 DBPR MCLs and MCLGs Regulated Contaminants MCL MCLG mg/L mg/L TTHM 0.080 -‐ Chloroform -‐ 0.07 Bromodichloromethane -‐ zero Dibromochloromethane -‐ 0.060 Bromoform -‐ zero HAA5 0.060 -‐ Monochloroacetic acid -‐ 0.070 Dichloroacetic acid -‐ zero Trichloroacetic acid -‐ 0.020 Bromoacetic acid -‐ -‐ Dibromoacetic acid -‐ -‐
The rule targets systems with the greatest risk and builds incrementally on
existing rules, aiming to decrease DBP exposure and related potential health risks and
provide more equitable public health protection (USEPA, 2015).
Three general strategies have been adopted to deal with DBP violations: (1)
switch from chlorination to an alternative disinfectant or disinfection regime, (2) reduce
DBP precursors in the raw water by enhanced treatment plant processes, and/or (3)
remove DBPs after they have formed. Although post-treatment or remote DBP control
has not received as much attention as the other two control strategies, i.e., switching
from chlorination and reducing organic precursors before the disinfection process,
remote DBP control has the potential to be a cost-effective treatment option and
compliance strategy (especially for small systems) compared to in-plant treatment
where all the water must be treated. Such cost effective compliance may prevent the
proliferation of chloramination, which does not meet the intent of the DBP rule by
3
forming unregulated DBPs (e.g., nitrosamines) some of which may be of more health
concern than the currently regulated THMs and HAA5s.
The overall goal of this project is to evaluate and model the use of granular
activated carbon for the control of preformed DBPs and DBP precursors.
1.2 Research Objectives
Use of granular activated carbon (GAC) in the distribution system would remove
both regulated classes of disinfection byproducts (DBPs) at the point of treatment, and
may lower DBP re-formation by removing DBP precursors, measured as total organic
carbon (TOC) without requiring significant investment in existing treatment or
disinfection facilities. While the adsorption capacity of GAC to remove regulated DBPs
is relatively low (McGuire et al, 1991; Tung et al., 2006), adsorption of DBPs may still
be economical in remote systems because only a small portion of the total system flow
must be treated. Alternatively, GAC can be used in a biological treatment mode to
degrade haloacetic acids (HAA’s) in steady state (Xie & Zhou, 2002). While this
approach will likely only reduce the regulated HAA’s, the treatment system will be able
to operate for long periods of time with very little maintenance. Remote GAC
adsorption/biodegradation would take place in an above ground pressure vessel, with a
schematic of the proposed treatment system shown in Figure 1-1.
Figure 1-1: Remote GAC treatment schematic
4
The project objective is to develop and evaluate the use of GAC in the distribution
system to meet DBPs regulations under both adsorptive and biological modes. It is
hypothesized that a post-treatment reactor strategically located in the distribution system
will offer small systems a cost-effective alternative to controlling THMs, HAA5s and
other unregulated DBPs. To verify our hypothesis, the following two primary research
questions will be answered:
1. How long can the GAC remove THMs and HAAs by adsorption under different
conditions?
2. What levels of HAA removal can be expected in a remote, engineered biological
treatment system under different conditions?
1.3 Thesis Organization
This thesis is divided into four chapters. Chapter 1 provides an introduction to the
material contained within. Chapter 2 is a literature review of the treatment methods
studied, providing a lens through which to view this work. Chapter 3 outlines the
materials and methods used throughout this research. Chapter 4 showcases experimental
results and discussion, with a summary of those results viewed in light of the research
objectives. Finally, appendices A - F are included which contain raw data, tables and
figures not shown in the body of the thesis.
5
Chapter 2 Background
2.1. Disinfection By Product Formation and Control
Chlorination is the most common disinfection method for drinking water.
Although chlorination is unquestionably important to the supply of safe drinking water,
chlorinated DBPs can be created through unintended reactions of chlorine with natural
organic matter (NOM), as well as bromide, (Eqn. 2-1). NOM is the principal precursor of
chlorinated DBPs in most water, and represents a significant portion of all organic matter
in most source waters (Singer, 1994).
𝐻𝑂𝐶𝑙 + 𝐵𝑟! + 𝑁𝑂𝑀 à 𝑇𝐻𝑀𝑠 ,𝐻𝐴𝐴𝑠 ,𝑎𝑛𝑑 𝑜𝑡ℎ𝑒𝑟 𝐷𝐵𝑃𝑠 Equation 2− 1
Toxicology studies have shown THMs, HAAs and other DBPs to be carcinogenic
or to cause adverse reproductive or developmental effects in laboratory animals, and a
large number of epidemiological studies have shown an association between the
consumption of chlorinated drinking water, or exposure to it, and bladder, colon and
rectal cancer in humans (Babi et al., 2007).
The best available technologies (BATs) recommended by the US Environmental
Protection Agency (USEPA) for the control of DBPs include (Wu, 2012):
• Enhanced Coagulation for precursor removal
6
• GAC 10 – Granular activated carbon filter beds with an empty-bed contact time
of 10 minutes based on average daily flow and a carbon reactivation frequency of
every 120 days
• Nanofiltration (NF) – Membrane molecular weight cutoff of 1000 Daltons or less
• Chloramination – for consecutive systems
One of the most effective and economical methods to control DBPs in
conventional WTPs is to remove precursors (organic material) before they react with
disinfectants. Much research on DBPs removal has been focused on NOM removal while
only a few results have been recently reported on the removal of DBPs after formation in
controlled experiments (Xie & Zhou, 2002; Tung et al., 2006).
2.2. Adsorption by Granular Activated Carbon
Adsorption by GAC is a well-studied treatment technique for the removal of
NOM, taste and odor compounds, and synthetic organic chemicals (SOCs) in drinking
water treatment (Crittenden et al., 2012; Sontheimer et al., 1988). In the adsorption
process, the adsorbent is defined as the solid media on which adsorption occurs (i.e.
GAC), and the adsorbate is the compound (or contaminant) that undergoes adsorption
onto the adsorbent (Crittenden et al., 2012). Activated carbon is a highly porous material,
providing a large surface area to which contaminants may effectively adsorb (Sontheimer
et al., 1988). Adsorption is a mass transfer operation in which adsorbate present in
aqueous solution is transported into the porous adsorbent grain by means of diffusion,
then adsorbed or accumulated on the inner surface of the adsorbent and thus removed
from the liquid (Crittenden et al., 2012; Sontheimer et al., 1988). Physical adsorption and
7
chemisorption (Table 2-1) are both adsorption phenomenon known to occur, with the key
differences summarized in Table 2-1 (Crittenden et al., 2012; Sontheimer et al., 1988)
Table 2-1: Comparison of Physical and Chemical Adsorption (adapted from Crittenden et al., 2012)
Parameter Physical Adsorption Chemisorption
Use for water treatment
Most common type of adsorption mechanism Rare in water treatment
Process speed Limited by mass transfer Variable
Type of bonding
Nonspecific binding mechanisms such as van der Waals forces, vapor condensation
Specific exchange of electrons, chemical bond at surface
Type of reaction Reversible, exothermic Typically nonreversible, exothermic
Heat of adsorption 4–40 kJ/mol >200 kJ/mol
While physical adsorption and chemisorption can be distinguished easily at their
extremes, some cases fall between the two, as a highly unequal sharing of electrons may
not be distinguishable from the high degree of distortion of an electron cloud that occurs
with physical adsorption (Sontheimer et al., 1988).
GAC treatment occurs in a specific unit operation referred to as a contactor
system or filter, with the active adsorption zone (top half of Figure 2-1) traveling
downward through the bed as treatment progresses, producing the effluent profile
concentration pictured in the bottom half of Figure 2-1 (DiGiano, 1983). Contactor unit
design variables include flow-rate and volume. Empty bed contact time (EBCT) is equal
to the volume of the contactor normalized by the flow rate, or the bed length normalized
by the velocity and in tandem with design flow rate, determines the amount of carbon
required in a contactor. Reducing the flow rate through the filter or increasing the
8
contactor volume (and corresponding mass of carbon) can increase EBCT, with longer
EBCTs delaying breakthrough and producing longer filter run times (DiGiano, 1983).
Typical EBCTs for water treatment applications range between 5 to 25 minutes.
Normalization of breakthrough data on a bed volume basis allows comparison of filters
performing at different EBCTs.
Figure 2-1: GAC Contactor Schematic: Idealized Adsorption Zone and Resulting Breakthrough (Noto, 2016) Removal effectiveness, and resulting breakthrough profile of a specific
contaminant, is constrained by physical and chemical factors related to the properties of
both the adsorbent and contaminant. Organic materials with high carbon contents such as
wood, lignite and coal are used to manufacture GAC, with GAC properties varying with
feedstock. A widely used metric for characterizing GAC is the iodine number, which
gives a good indication of the microporosity of the GAC sample (Sontheimer et al. 1988).
Iodine numbers for the GAC utilized in this study are presented in Chapter 3, Material
and Methods. Adsorbability and a literature review of TOC, THMs and HAAs removal is
discussed in the following sections of this chapter.
9
2.2.1. TOC Adsorption
Roberts and Summers (1982), Babi et al., (2007), Johnson et al., (2009) and
Summers et al. (2010) studied TOC adsorption in GAC filters. They report 10 to 20 %
immediate breakthrough or nonadsorbable fraction of the TOC followed by a
breakthrough of different adsorbable fractions to a steady-state condition dominated by
biological removal.
Figure 2-2: Representative DOC breakthrough for activated carbon columns (Summers et al., 2010) Roberts and Summers (1982) reported that in most cases a nearly constant
concentration between 50 and 90 percent (mean of 80%) of the influent DOC appears in
the effluent after exhaustion of the GAC. Displacement of poorly adsorbable organics by
more strongly adsorbing compounds, biodegradation, and slow diffusion of humic
10
substances into the microporous carbon are cited as contributing factors to this behavior
(Roberts & Summers, 1982).
2.2.2. THM Adsorption
The literature indicates that adsorption capacity of GAC for trihalomethanes
varies widely depending on source water quality and application type. The Freundlich
equation (Eqn. 2-2) is often used to model the equilibrium adsorption capacity of
activated carbon. In equation 2-2, q is the solid phase adsorption capacity, C is the liquid
phase concentration and the Freundlich constants are K and n.
𝑞 = 𝐾 ∗ 𝐶!! Equation 2− 2
Table 2-2 lists the THM compound properties that affect adsorption affinity and
GAC capacity, including molar mass, octanol-water partition coefficient, solid phase
adsorption capacity at a liquid phase concentration of 10 µg/L, q10, and Freundlich
modeling parameters for adsorption on bituminous carbon.
Table 2-2: Trihalomethane Adsorption Affinity Indicators for Bituminous based GAC (Speth & Miltner, 1990; World Health Organization, 2004) Compound Molar Mass log Kow K 1/n q10 g/mol (mg/g)*(L/mg)^(1/n) mg/g TCM 119.37 1.97 9.4 0.67 0.43 DCBM 163.8 1.88 22.2 0.66 1.09 DBCM 208.28 2.08 47.3 0.64 2.53 TBM 252.73 2.38 91.8 0.67 4.30
THM compound properties that affect adsorption affinity include molar mass and
solubility, measured by the octanol-water partition coefficient. The octanol-water
11
partition coefficient (Kow) is the ratio of a chemical's concentration in octanol to its
concentration in the aqueous phase of a two-phase system at equilibrium. Increasing Kow
values indicate increasing hydrophobicity, and correspondingly, increasing affinity for
adsorption (McCarty et al., 1987). The adsorbability of the TTHM species is TCM à
DCBM à DBCM à TBM. This order of breakthrough has also been shown in columns
(Fokken & Kurtz, 1984). In adsorption isotherm results, chlorinated THM species gave
lower adsorption capacites (K) for GAC than their brominated analogues did (Speth &
Miltner, 1990).
When applied in a GAC column, the capacity for TCM (typically the THM
species with the highest concentration) is exhausted in a matter of weeks to months
(Table 2-3), while GAC may last months to years for TBM. Factors that impact the
effectiveness of GAC for treatment of THMs include adsorber EBCT, influent speciation
of THMs, carbon type utilized, competition for adsorption sites by NOM and other
contaminants, preloading of organics onto the carbon, temperature, pH, and adsorption
kinetics ,affected by carbon size and hydraulic loading rate (Speth & Miltner, 1990;
Johnson, et al., 2009).
The volume of water treated can be normalized to the volume of GAC in the
column and expressed as throughput in bed volumes (BV). The BV treated when C/C0
reaches 0.1 and 0.5 are referred to herein as “BV10” and “BV50” respectively, and are
used in comparing removal performance of a compound under different conditions. “Peak
C/C0” refers to the maximum chromatographic effect (normalized concentration greater
than one) reported in that study.
12
Table 2-3a: THM Breakthrough Literature Review
Reference Properties Influent Water Characteristics Study Specifics Breakthrough Profile
Title Compound Cl2 TOC0 THM C0 EBCT(s) Loading Rate Scale GAC BV10 BV50
Peak C/C0
mg/L mg/L μg/L min m/h BV x 103
Babi et al., (2007) TTHM 0.50
Avg: 2.0 mg/L,
range: 1-‐5mg/L
avg: 60 , range (20-‐
170) 14 min
4.8 m/h (range 4–6 m/h);
Pilot
Filtrasorb F-‐400,
Chemviron Carbon, 12x40
mesh
5 16 4
Kim & Kang, (2008) TTHM
0.69 ± 0.49 mg/L
2.7 ( ± 0.6) 161 ± 54 9.8 min 4.58 Full-‐
Scale
Calgon F 820 -‐ Bituminous
Coal -‐ 14 -‐
Johnson et al., (2009)
TTHM
1.06 2.5
avg 73, range (60 -‐
95)
10 -‐ Pilot Calgon F 600 -‐ Bituminous
Coal
10 17 0.8
TCM 27 6 11 1.6
DCBM 23.5 13 20 -‐
DBCM 19 21 NBT -‐
TBM 3.5 NBT NBT -‐
Corwin & Summers, (2010)
TCM -‐ 2.7 70 7 -‐ RSSCT Calgon F300 -‐ Bituminous 25 30 -‐
Fokken & Kurtz, (1984)
TCM
-‐ 1.1
0.64
8 10 -‐ Row 0.8S
7.5 15 1.7
DCBM 1.6 12.5 21 1.1
DBCM 2.8 15 25 1.25
TBM 2.9 22 35 1
Sontheimer et al., (1988)
THM -‐ 3 6.4 15 -‐ -‐ F 300 3 5.5 1.5
Meijers, A.P., et al., (1984)*
TCM
0.8 2-‐4
30
12 Full-‐Scale
Norit Row 0.8 Supra
2 7 -‐
DCBM 25 3 11 -‐
DBCM 12 5 15 -‐
Meijers, A.P., et al., (1984)*
TTHM -‐ 4.5 avg: 70 , range(30-‐
130)
15 12 Full-‐Scale
Norit Row 0.8 Supra
2 3.5 1.1
TTHM 30 12 3 11 -‐
*Indicates symposium papers compiled in NATO Committee on the Challenges of Modern Society, 1984
13
Table 2-3b: THM Breakthrough Literature Review
DeMarco & Brodtmann., (1984)*
TCM -‐ 4
Avg: 5 , range (3-‐33.5)
16.3 2.2 Pilot
WVG 12
3.1 4.5 2
13.6 2.7 Full-‐Scale 3.1 4.7 1.1
Avg: 7 , range (3-‐47)
21.4 2.2. Pilot 2.3 3.7 4
17.5 2.7 Full-‐Scale 2.9 4.5 2
10.9
5.34 Pilot
3 5 4
21.8 -‐ -‐ -‐
32.7 3.3 5 1.1
43.6 NBT NBT -‐
Wood, P., & DeMarco, J., (1984)*
TCM 3 6.4 Avg: 67.3 , range (45-‐
131) 6.2 7.33 Bench
-‐Scale
Nuchar WVG, Westvaco 3.2 5.8 2
Hydrodarco 1030, ICI Americas
3.2 5.8 1.3
Filtrasorb F400 , Calgon 3.8 6.5 1.6
Witcarb Grade W950 ,
Witco 6.5 9.2 -‐
Miller, R. (1984)* TTHM -‐ 2
Avg: 40 , range (10-‐
75)
4.5
6.1 Pilot
Bituminous 12x40
4.2 9.5 -‐
7.5 4.8 15.3 -‐
7.5 Bituminous 20x50
4.8 16.3 -‐
*Indicates symposium papers compiled in NATO Committee on the Challenges of Modern Society, 1984
14
Desorption due to competitive adsorption and concentration gradient reversal has
been shown to cause chromatographic peaking in many studies (Babi et al., 2007;
Johnson et al., 2009; Sontheimer et al., 1988).
Sontheimer et al. (1988) reported a reduction in micropollutant adsorption
capacity in columns preloaded with NOM, but difficulty predicting the fouling effect of
NOM in columns due to most natural waters having different concentrations and types of
humic substances. On-site pilot plant studies give the best results for evaluating the
impact of NOM on adsorption due to the variability source water quality and level of
pretreatment (Babi et al., 2007). The impact of TOC on GAC adsorption capacity of
TTHM from available literature values is displayed in Figure 2-3. Variability in this data
is due to the different carbon types, levels of pretreatment, and scales of the various
studies in Table 2-3. Higher influent TOC significantly shortens filter run time (bed
volumes) to 50% breakthrough.
15
Figure 2-3: TTHM BV50 as a function of influent TOC concentration for GAC columns with the influent TTHM concentration greater than 10 µg/L GAC type has significant impacts on adsorption of THMs. Coconutd based GACs
have the highest iodine numbers, which correspond to a higher capacity to adsorb small
molecules, such as volatile organic chemicals (Sontheimer et al., 1988).
The literature is unclear with regards to the effect of EBCT on adsorption of
THMs in GAC. Generalizing to micropollutants in the microgram/L range, optimal
carbon utilization (specific throughput) has been shown at shorter EBCTs (10 min) due to
fouling of the carbon by NOM for longer EBCTs (Sontheimer et al. 1988). Smaller
EBCTs give shorter running times until micropollutant breakthrough and hence, less time
for carbon fouling to occur (Sontheimer et al. 1988). Better THM removal on a bed
volume basis is then expected for shorter EBCTs.
y = -‐1.7336x + 17.164 R² = 0.52208
0
2
4
6
8
10
12
14
16
18
0 2 4 6 8 10
TTHM
BV 5
0 x 103
Influent TOC (mg/L)
TTHM >10 μg/L
16
Significant gaps in the literature exist with regards to speciated data for THM
removal under varying influent conditions and EBCTs. This research aims to fill those
gaps by producing speciated breakthrough data for a variety of influent TOC, Br and Cl2
conditions.
2.2.3. HAA Adsorption
Studies by Tung et al. (2006) and Xie and Zhou (2002) have indicated that that
the GAC adsorption capacities for some HAAs were much lower than for those for
THMs, with TCAA being the exception. Adsorption studies conducted by Liu and
Andrews (2001) and Speth and Miltner (1998) indicated that HAA species having a
higher halogen number gave a larger adsorption capacity (K) for GAC (Tung et al.,
2006). In adsorption isotherm results, chlorinated HAA species had lower adsorption
capacities (K) for GAC compared to their brominated analogues (Speth & Miltner, 1990).
Full scale and laboratory GAC filter studies have shown high levels (>90%) of HAA
adsorption to occur for as short as eight days to as long as three months before 50%
breakthrough of HAA5 (Liu et al., 2001; Xie & Zhou, 2002; Kim & Kang, 2008).
2.3. Biological Activity in GAC Filters
Biomass has been shown to develop in filters both with and without disinfectant
free chlorine residual (Xie & Zhou, 2002; Wu & Xie, 2005; Chuang et al., 2011; Zearley
& Summers, 2012;). As water percolates through the filter bed natural occurring
heterotrophic bacteria attached to the filter medium (e.g. GAC) oxidize organic matter for
energy supply and carbon source.
17
In most drinking water biofilters, the primary substrate sustaining the microbial
biomass is the biodegradable fraction of the dissolved organic matter (DOM) measured as
TOC. Primary substrate must occur at concentrations above a threshold concentration
(Smin) needed to support primary cellular processes without another substrate present
(Zearley & Summers, 2012). Micropollutants such as THMs and HAAs are classified as
secondary substrates, present below concentration Smin, and are removed by secondary
substrate utilization or cometabolism (Zearley & Summers, 2012). The research of Zearley and Summers (2012) showed a range of trace organic contaminants to follow a
pseudo-first order rate model, with removal efficiency independent of influent
concentration. The contaminant utlitization rate constant and biomass can be represented
by a pseudo-first order rate constant, k’.
2.3.1 TOC Biodegradation
Primary substrate utilization has been represented by TOC removal across
biofilters since biodegradation is the only significant removal mechanism of DOM with
non-adsorptive media (Zearley & Summers, 2012). Exhausted GAC is assumed to be
non-adsorptive, with steady state removal of TOC in the range of 2 -20% reported in
studies by Babi et al., (2007), Kim and Kang (2008), Johnson et al., (2009) and Zearley
and Summers (2012).
2.3.2 THM Biodegradation
Aerobic biodegradation of THMs in GAC columns is not thermodynamically
favorable due to their high oxidation states (Kim & Kang, 2008; Babi et al., 2007).
18
2.3.3 HAA Biodegradation
High levels of HAA biodegradation has been reported in GAC biofilter studies,
with typical removals for established steady state systems exceeding 90% for HAA5
(Kim & Kang, 2008; Tung et al., 2006; Johnson et al., 2009; Wu & Xie, 2005). A
summary of the results of past GAC column studies is presented in Table 2-4.
19
Table 2-4: HAA Biodegradation Literature Review
Reference Properties Influent Water Characteristics Study Specifics Removal and Acclimation
Title Compound Cl2 TOC0 HAA C0 Scale Loading Rate EBCT(s) Temp
Steady-‐State Removal
Time to Steady State
mg/L mg/L μg/L m/h min °C % Babi et al., (2007)
HAA5 0.50 2 Pilot 4.8 m/h 14 15 >90 Unable to discern
due to adsorption
Kim& Kang, (2008)*
HAA5 0.69 2.7 205 ( ± 98) Full-‐Scale 4.58 9.8 5 34
6 months 23 99
Tung et al., (2006)
ClAA -‐ -‐
2.0 Full-‐Scale 3.42 10 -‐
100 30 days
Cl2AA 25.0 95 50 days
Johnson et al., (2009)
HAA5 1 2.5 25 Pilot-‐Scale -‐ 10 12-‐18 100 7 months
Zhou & Xie, (2002)
ClAA
1-‐2 -‐
50
Bench-‐Scale -‐ 20 20-‐22
100 35 days
BrAA 50 100 50 days
Cl2AA 50 100 70 days
Br2AA 50 100 70 days
Cl3AA 50 100 Unable to discern
due to adsorption
HAA5 250 100 70 days
Wu & Xie, (2005)**
HAA6 1-‐2 -‐ 300
5.5 5
4 28
10 58
20 95
30 100
2.8 10
4 52 Media collected from GAC filters that had been online for 2.5-‐3
years
10 85
20 98
Bench-‐Scale 30 100
1.9 15
4 70
10 95
20 100
30 100
1.4 20
4 90
10 98
20 100
30 100
*Speciated DCAA and TCAA data available in report
**Full speciated EBCT, Temperature and rate constant data for ClAA, Cl2AA, BrAA, BrClAA, Br2AA and Cl3AA available in report
20
The biodegradability of the HAA species in a drinking water biofilm is MBAA >
MCAA > BCAA > DCAA > DBAA > TCAA (Bayless & Andrews, 2008). Di-
halogenated species were removed to a lesser extent than the mono-halogenated
compounds, with the results of Zhou and Xie (2002), Baribeau et al. (2005), Kim and
Kang (2008), and Chuang et al. (2011) showing that DCAA is more biodegradable than
TCAA. Wu and Xie (2005) and Kim and Kang (2008) have reported significant effects of
temperature and EBCT on HAA biodegradation, with higher temperatures and EBCTs
corresponding with higher levels of removal due to biodegradation. The HAA5 biofilter
results from six studies are shown in Figure 2-4 and illustrate the impact of EBCT and
temperature.
Figure 2-4: Effect of Temperature and EBCT on HAA5 Biodegradation (data from six references in Table 2-4)
y = 29.686ln(x) + 12.636 R² = 0.96125
y = 3.9176x + 11.706 R² = 0.99253
0
20
40
60
80
100
0 5 10 15 20 25
Removal (%
)
EBCT (min)
T > 15 C
T 6-‐10 C
T < 5 C
21
Kim and Kang (2008) reported a decrease from 99% HAA5 removal in summer
months to 34% HAA5 removal in winter months. The kinetic analysis of Wu and Xie
(2005) shows that HAA degradation rates increase at higher temperatures, with high
removal rates at colder temperatures being obtained by significantly increasing EBCT.
22
Chapter 3 Materials and Methods
3.1 Materials
3.1.1 Activated Carbon Specifications
Three types of granular activated carbon (Calgon F400, Norit HD 4000,
AquaCarb 1230C) were used in adsorptive mode RSSCTs and one type of carbon
(AquaCarb 820) was used in pilot scale biodegradation experiments. The properties of all
carbons as received from the manufacturers are summarized in Table 3-1.
Table 3-1: Granular Activated Carbon Manufacturer Specifications
Carbon Type ID
Mesh Size
Min Iodine
No.
Effective Size
Uniformity Coefficient
Apparent Bed Density
Abrasion No.
Moisture (max)
U.S. Sieve mg I2/g mm max g/cm3 Wt.% Wt. %
Bituminous Coal
Calgon F400 12 x 40 >1000 0.55-
0.75 1.9 0.54 75 2
Lignite Coal
Hydrodarco 4000 10 x 30 >500 0.6-0.8 2.1 0.39 70 8
Coconut Shell
AquaCarb 1230C 12 x 30 1100 0.6-0.85 2.0 0.46-0.52 85 -‐
Bituminous Coal
AquaCarb 820 8 x 20 900 1.0-1.2 1.5 0.46-0.54 80 -‐
For use in RSSCTs, the carbons were carefully crushed with a mortar and pestle and
separated with US Standard sieves on a sieve shaker. The fractions between the the #100 and
#200 sieves (dp=0.11 mm) were collected for bench-scale experiments. The crushed GAC
fractions were washed, dried, and stored in a desiccator until use (EPA, 1996).
23
Media from the Southern Nevada Water Authority’s (SNWA) River Mountains
water treatment facility was shipped to the University of Colorado for utilization in the
biofiltration pilot study. The full-size bituminous AquaCarb 820 GAC had previously
been in contact with a chlorine residual between 1.5 and 2 mg/L Cl2 for more than 5
years. Initial biomass activity was 11,000 pg ATP/g. A baseline measurement for a
carbon with no biomass would be expected to be 0 pg ATP/g, with details of the total
ATP Luminultra method located in Appendix H.
3.1.2 Source Waters
Two source waters were used in this study, with various chemical amendments
and mixtures used to simulate various influent conditions. The typical measured ranges of
the source waters are summarized in Table 3-2.
Table 3-2: Source Water Quality
Source Water
DOC pH Alkalinity as CaCO3
UVA SUVA
mg/L -‐ mg/L cm-‐1 (L mg-‐1 m-‐1)
Boulder Tap 1.3-‐2.2 7.9 40* 0.015 -‐ 0.026 1.15 -‐ 1.18
Wonderland Lake 9.88 8.4 120 0.159 1.6
*from past research
3.1.3 Chemicals
Laboratory grade 5.65-6% sodium hypochlorite solution (CAS 7681-52-9, Fisher
Scientific) and potassium bromide salt (CAS 7758-02-3, Fisher Scientific) were dosed
into the source waters to provide additional exposure to chlorine and bromide for DBP
24
formation. Dosed waters were held for a minimum of 24 hours to provide ample time for
formation.
3.2 Methods
3.2.1 Analytical Lab Methods
Table 3-3: Analytical Methods
Analyte Measuring Units
Detection Limit Equipment/Procedure Reference method
pH/Temp N/A N/A Denver Instruments Model 220 pH and conductivity meter
SM 4500-H+
TOC/DOC ppb 4 Sievers 5310 C TOC SM 5310 C
UVA cm-1 0.001 Hach DR-4000 UV Spectrophotometer SM 5910 B
Alkalinity mg/L as CaCO3
2 Hach Digital Titrator Model 16900-01 SM 2320 B
Free chlorine mg/L as Cl2
0.02 Hach Pocket Colorimeter/Hach Method 8021
SM 4500-Cl G
Total ATP (tATP) pg/g -
Lumitester™ C-110 Luminometer & Equipment Set (EQP-PAC-C110) Deposit & Surface Analysis
-
Chloroform µg/L 0.82 Agilent 6890 GC EPA Method 551.1 Dichlorobromomethane µg/L 0.37 Agilent 6890 GC EPA Method 551.1 Chlorodibromomethane µg/L 0.32 Agilent 6890 GC EPA Method 551.1 Bromoform µg/L 0.34 Agilent 6890 GC EPA Method 551.1 Chloroacetic Acid µg/L 0.95 Agilent 6890 GC EPA Method 552.2 Bromoacetic Acid µg/L 0.87 Agilent 6890 GC EPA Method 552.2 Dichloroacetic Acid µg/L 0.96 Agilent 6890 GC EPA Method 552.2 Trichloroacetic Acid µg/L 0.84 Agilent 6890 GC EPA Method 552.2 Dibromoacetic Acid µg/L 0.91 Agilent 6890 GC EPA Method 552.2
25
A linear relationship (R2=0.91) between two common biomass analysis techniques, the
Luminultra total ATP method and a phospholipids based method, has been shown when
applied to media from similar source (Dowdell & Summers, 2012). The total ATP
method was used in this research, as the phospholipid method is very time intensive.
3.2.2 Rapid Small Scale Column Tests (after Kempisty, 2014)
The rapid small scale column test (RSSCT) was used for all of the adsorptive
mode GAC experiments in this project. Variables that were modified include GAC type,
DBP speciation, and EBCT while maintaining the same general design.
The EPA Manual for Bench- and Pilot-Scale Treatment Studies guided the set-up
of the RSSCTs (EPA, 1996). Figure 3-1 shows a generic diagram of the RSSCT setup.
Tap water was transferred to 55 gallon HDPE barrels and either left unammended, or
spiked with sodium hypochlorite or potassium bromide to create the desired influent
condition. After being left at lab temperature (21°C) for 24 hours to allow DBP
formation, the barrels were transported to a walk in refrigerator and stored at 4°C.
26
Figure 3-1: Base RSSCT Set Up (after Kempisty 2014)
The refrigerated water was then transferred to smaller HDPE carboys and brought
to room temperature as needed for the RSSCT feed. Tubing consisted of 4.76 mL
polytetrafluoroethylene (PTFE or Teflon) or 1/4” outer-diameter stainless steel tubing
(Nalgene 890 FEP by Thermo Fisher Scientific Inc., Waltham, MA). Valves and fittings
were manufactured by Swagelock (Solon, OH). All pumps were PTFE diaphragm pumps
made by Cole-Parmer (Vernon Hills, IL) with diaphragm model 7090-62. Two different
drives were used, model numbers 77521-50 and 7521-40.
Other materials used were 5 gallon plastic carboys for effluent collection, pipettes
(Eppendorf International, Hamburg, Germany), and glass wool. The glass wool was used
27
as support for the GAC adsorbers inside of the PFTE columns and also as a prefilter. The
prefiltration acted to remove any particulate matter that could cause a blockage of flow
which would cause a pressure increase to the point where the pump could not move water
through the columns.
There were two GAC columns in series during most experiments, with one
RSSCT “BTBr” having three GAC columns in series. The first two columns
corresponded to 5 minute EBCTs yielding an overall EBCT of 10 minutes for most
experiments. For the “BTBr” influent water RSSCT, a third column corresponding to at
10 minute EBCT was added to the first two, yielding an overall 20 minute EBCT. Valves
were used between the columns to allow sampling at 5 minute, 10 minute and 20 minute
EBCTs at the correct flow rate. The columns were created by pushing a glass wool plug
as a base for the GAC into the bottom of a 4.76 mm diameter column. The ground GAC
was added using Pasteur pipettes to the column, already full of DI water. After each
addition of ground GAC, the column was gently rapped with a wrench to ensure settling
of the carbon. This was important because the volume of GAC was used to determine the
correct amount of contact time.
The glass wool prefilters were changed every 7-14 days depending on visual
inspection and system pressure. A pressure gauge was installed before the columns to
measure the pressure to determine if clogging of the GAC was occurring. A pressure
dampener was installed before the columns to moderate the flow to a steady level instead
of the pulsing created by the diaphragm pump. Effluent was collected in a plastic 5-
gallon carboy.
28
Influent samples were taken when new batches of water were created in 55 gallon
barrels. Effluent TOC samples were collected every 1-3 days. At the same time, the
runtime between samples and effluent volume was measured and used to calculate flow
rate and overall throughput. Throughput was reported in terms of bed volumes of the
column. One bed volume of water equals the volume of the GAC in the column. Another
way to report the amount of water treated is in terms of the ratio of GAC mass to the
volume of water treated. This is expressed as the carbon use rate (CUR). The CUR allows
direct comparison of amount of utilized carbon per volume of water treated, making it a
good measure for utilities. The calculated CUR is defined as the density of the GAC
divided by the bed volumes of throughput.
The RSSCT is based upon using GAC of a smaller diameter and maintaining
similitude of dimensionless parameters so that the RSSCT will behave like a full-size
adsorber. An RSSCT designed using a scaling factor and can be used to replicate the full-
scale data in as little as 4% of the time as a pilot scale study (Crittenden et al., 1986a).
Crittenden, et al. (1986a) showed that the EBCTs of an RSSCT and full-scale
adsorber can be related to the particle sizes and intraparticle diffusivity of each adsorber,
as shown in Equation 3-1.
𝐸𝐵𝐶𝑇!"𝐸𝐵𝐶𝑇!"
=𝑅!"𝑅!"
!
∙𝐷!"𝐷!"
Equation 3− 1
The radius of the GAC is represented by R and the intraparticle diffusivity is
represented by D. It does not matter if the radius or diameter is used, but diameter can be
more convenient to work with because activated carbon vendors and sieves generally use
29
diameter to report size. Equation 3-2 defines the scaling factor, or the proportion that is
used to relate the large column (LC) and small columns (SC), mathematically and
therefore the design.
𝑆𝐹 =𝑅!"𝑅!"
Equation 3− 2
The scaling factor, or ratio of particle diameters, for all of the RSSCTs in this
study was 8.5. The Proportional Diffusivity (PD) RSSCT approach assumes that the
diffusivities are linearly proportional to the particle size, so Equation 3-1 becomes the
design equation for a PD-RSSCT, Equation 3-3.
𝐸𝐵𝐶𝑇!"𝐸𝐵𝐶𝑇!"
=𝑅!"𝑅!"
Equation 3− 3
The scaling equations recently developed by Corwin and Summers (2012) and
Kempisty (2014) to improve prediction of full-scale GAC capacity are justification
supporting the use of the PD-RSSCT approach in this research.
One way to relate RSSCT performance with a theoretical full-scale adsorber is the
full-scale operating time (FSOT). The FSOT is calculated as the ratio of the volume of
water that has passed through the RSSCT to the volume of the bed, and using the EBCT
of the columns, shown in Equation 3-4.
𝐹𝑆𝑂𝑇 = 𝐵𝑉𝑠 ∙ 𝐸𝐵𝐶𝑇 =𝑉𝑜𝑙𝑢𝑚𝑒!"#$%𝑉𝑜𝑙𝑢𝑚𝑒!"#$%&
Equation 3− 4
30
The scaling factor is used to calculate the FSOT because the volume of the bed is
based upon the length, and the length is calculated by dividing the EBCT divided by the
scaling factor, shown in Equation 3-5.
𝐿𝑒𝑛𝑔𝑡ℎ!" =𝐸𝐵𝐶𝑇𝑆𝐹 Equation 3− 5
The scaling factor is the basis of the calculation to determine the size of the
RSSCT and also defines the relation to full-scale adsorbers.
3.2.3 Fixed-‐Bed Adsorption Modeling (after Kempisty, 2014)
Adsorption Design Software (AdDesignS) from Michigan Technological
University offers three different models to predict target organic removal using GAC
including the Equilibrium Column Model (ECM), the Constant Pattern Homogeneous
Surface Diffusion Model (CPHSDM) and the Pore and Surface Diffusion model (PSDM)
(Kempisty, 2014). The PSDM is a mechanistic model of fixed bed adsorption that has
been shown to successfully model multi-solute adsorption systems, and was exclusively
used in this modeling effort. The PSDM requires input of the system design and
operating parameters such as, particle diameter, bed porosity, bed density, EBCT, filter
approach velocity, and initial concentration of the target compound. The Freundlich
isotherm parameters K, and 1/n, film mass transfer coefficient, tortuosity, and the surface
and pore diffusion coefficients are also required (Corwin & Summers, 2010). Corwin
demonstrated that intraparticle diffusion is responsible for the majority of mass transfer
control in typical water treatment plants (Corwin & Summers, 2012). Other work has
shown that in the presence of DOM, intraparticle diffusion is dominated by pore diffusion
and surface diffusion can be considered negligible (Kempisty, 2014). Further discussion
31
of the PSDM model inputs and their impact is presented in the dissertations of Corwin
and Kempisty who both extensively modeled micropollutant breakthrough using the
PSDM (Corwin, 2012; Kempisty, 2014).
Performing a total of four RSSCTs on bituminous carbon under varying
conditions produced enough data to explore modeling implications. Of particular interest
are competitive adsorption effects, and this modeling effort aims to quantify whether
NOM – THM interactions, THM-THM interactions, or a combination of both are
controlling adsorption of THMs on bituminous GAC.
Kempisty (2014) showed that GAC adsorption capacity for cVOCs was
negatively affected by both DOM and co-solute competition. It has been shown by
modeling and experimentation that compounds of similar adsorption strength tend to
compete for adsorption sites more strongly than compounds of differing adsorption
strengths (Kempisty, 2014). When there are multiple co-solutes (in our case TCM,
DCBM, DBCM, TBM), competition for adsorption sites on the activated carbon is
expected. It has also been shown that in the presence of TOC, bed volumes to 10%
breakthrough of VOC’s were reduced by 28% when comparing the low-TOC water
(TOC=0.3 mg/L) against organic-free water, with larger differences observed for higher
TOC waters (Kempisty, 2014). Sontheimer et al., 1988 reported 36 – 86 % capacity loss
for chloroform from preloading a carbon with tap water. NOM-solute model runs are
based on empirical relationships developed using waters containing varying influent
characteristics.
32
3.2.4 Biofilter Pilot Column Tests (after Zearley, 2012)
The biofilters were packed into 25 mm inner diameter laboratory glass columns
(ACE Glass 5820-37) with Teflon end caps and stainless steel fittings. Every filter had a
layer of support media (2 mm glass beads) below the filter media. The support media was
not included in the calculation of the EBCT. A needle valve after each column was used
to control flow. Sampling ports were located immediately before and after each column
to assess the removal associated directly with the filter. The biofilters were gravity fed
from multiple HDPE feed barrels located in an upstairs laboratory. The feed barrels were
refilled as needed, usually every two to three days.
Three biofilter setups were operated in parallel as a one-pass system to simulate
full-scale operation. Each biofilter setup consisted of three columns in series (Figure 3-2)
with a target overall EBCT of 20 min. The target hydraulic loading rate (HLR) for all of
the filters was 4.14 m hr-1. While these loading rates are on the low end of filter operation
rates, they facilitated the operation of the filters as they decreased the required volume of
water. All of the systems were operated at lab temperature (20 ± 2 °C), which is within
the range of temperatures that, depending on geographic location, most water treatment
facilities experience. The flow varied due to biomass and particle buildup within the filter
and was measured every 2 to 3 days and adjusted as needed. The change in hydraulic
head due to the water level decreasing in the feed tanks did not cause a measurable
change in the biofilter flow rate. The flow was monitored by measuring the amount of
water collected in a graduated cylinder in 1 min and the flow was adjusted by a needle
valve immediately after the biofilter.
33
Figure 3-2: Biofilter Setup
Influent and effluent samples were collected from sampling ports immediately
before and after each biofilter. The day prior to sampling, the flow was measured, and if
required, adjusted to the target hydraulic loading rate. The flow was rechecked, and
adjusted as needed, prior to sampling. If adjusted, a minimum of 10 bed volumes were
allowed to pass before samples were taken.
The biofilters were sampled for TOC, THMs and HAAs approximately once per
week for the duration of each month long run, for a total of 3 sampling events per run.
The biofilters were operated for two months total, with the first month long run
investigating the effect of influent bromide and the second month long run investigating
the effect of influent TOC. Paired influent and effluent samples were taken at all times.
34
Chapter 4 Results and Discussion
Experiments utilizing RSSCTs to assess adsorption behavior and pilot scale
biofilters to assess biodegradation behavior were carried out. Results from the adsorption
columns are presented in Sections 4.1 and 4.2 and those from the biofilters in Section 4.3.
Two RSSCTs were performed. RSSCT #1 evaluated the effect of three different GAC
types on TOC and DBP removal, with one of the carbons moving on to further testing in
RSSCT #2. RSSCT #2 evaluated the impact of source water quality on GAC filter
performance using the selected carbon from RSSCT #1. Results from both RSSCTs and
a modeling effort using the Pore Surface Diffusion Model (PSDM) are discussed. Pilot
scale biofilters were operated for a period of about two months, with the first month
(Phase 1) investigating the impact of influent bromide on HAA biodegradation, and the
second month (Phase 2) investigating the impact of influent TOC and temperature on
HAA biodegradation.
4.1 Effect of GAC type (RSSCT #1)
A set of three RSSCTs with GAC from three different base materials were run to
evaluate the effectiveness of GAC for THM adsorption. Bituminous-based, lignite-based
and coconut-based activated carbons were evaluated in RSSCT #1 for TOC and THM
removal, with the experimental results used to choose a GAC type to test in further
RSSCT and pilot systems under differing influent, EBCT, and temperature conditions.
35
In a GAC bed, once the mass transfer zone reaches the end of the bed, target
compounds begin to appear in the effluent. The effluent concentration can be expressed
as a normalized effluent (C/C0), defined as the ratio of the effluent concentration to the
influent concentration.
In this section, breakthrough results for TOC and THM are presented and
discussed. The same influent water was supplied to all three carbons in order to compare
performance and the influent water quality is summarized in Table 4-1. For each RSSCT
set up, Boulder tap water was supplied in batches from a 40L Nalgene container for
between 8 to 12 days.
Table 4-1: RSSCT #1 Influent Characteristics
TOC TTHM TCM DCBM DBCM TBM pH mg/L µg/L µg/L µg/L µg/L µg/L 1.3 29 28.2 0.9 BDL BDL 7.9
BDL – below detection limit of 0.3 µg/L
4.1.1 TOC Adsorption
TOC adsorption is important to this study, as low TOC removal would result in
reformation of high levels of DBPs upon rechlorination. Rechlorination in the distribution
system post GAC treatment must occur as GAC reacts with the chlorine and a chlorine
residual is required at all points in the distribution system, termed secondary disinfection.
Thus, a GAC that effectively removes both TOC and THMs is most desirable. Effluent
TOC was sampled at 5 and 10 minute EBCTs, with the results shown in Figure 4-1.
Previous studies have shown THM breakthrough to lag behind TOC breakthrough in
columns designed for micropollutant removal (Sontheimer et al., 1988; Johnson et al.,
2009).
36
Figure 4-1: TOC Breakthrough at 5min EBCT for three different GAC types - (Inf. TOC = 1.3 mg/L)
37
Figure 4-2: TOC Breakthrough at 10min EBCT for three different GAC types - (Inf. TOC= 1.3 mg/L)
The effect of carbon type on TOC breakthrough at both 5 and 10 minute EBCTs
is shown in Figures 4-1 and 4-2. In all columns a non-adsorbable fraction of 10 to 15 %
was observed. The lignite and bituminous GACs showed similar behavior with BV50
values of 14,000 at both EBCTs, while the BV50 for the coconut GAC at both EBCTs
was about 3,000. TOC breakthrough for the bituminous GAC is similar to that predicted
by the Zachman and Summers (2010) model which predicts 50% breakthrough at about
16,000 BV. Coconut carbons are known to have more micro porous pore structure than
their coal-based counterparts, and have been shown to perform poorly for TOC removal
(Palmdale Water District, 2011; Potwara, 2012).
38
4.1.2 DBP Removal
THM compound properties that affect adsorption affinity and GAC capacity,
including molar mass, octanol-water partition coefficient, solid phase adsorption capacity
at an arbitrary liquid phase concentration of 10 µg/L, and Freundlich modeling
parameters for adsorption on bituminous carbon are shown in Table 4-2. The
adsorbability of the TTHM species is TCM à DCBM à DBCM à TBM. This order of
breakthrough has also been shown in columns (Fokken & Kurtz, 1984).
Table 4-2: Trihalomethane Adsorption Affinity Indicators for Bituminous based GAC (Speth & Miltner, 1990; World Health Organization, 2004) Compound Molar Mass log Kow K 1/n q10 g/mol (mg/g)*(L/mg)^(1/n) mg/g TCM 119.37 1.97 9.4 0.67 0.43 DCBM 163.8 1.88 22.2 0.66 1.09 DBCM 208.28 2.08 47.3 0.64 2.53 TBM 252.73 2.38 91.8 0.67 4.30
The octanol-water partition coefficient (Kow) is the ratio of a chemical's
concentration in octanol to its concentration in the aqueous phase of a two-phase system
at equilibrium. Increasing Kow values indicate increasing hydrophobicity, and
correspondingly, increasing affinity for adsorption (McCarty et al., 1987). The
experimentally determined Freundlich isotherm parameter “K” is an indicator of
adsorption capacity used in modeling. The amount of solute adsorbed per unit weight of
adsorbent is proportional to “K” and hence, increasing values of “K” indicate increasing
adsorbability (Sontheimer et al., 1988).
HAA adsorption results for adsorptive RSSCT #1, listed in Appendix D, were
nonsystematic and thus not analyzed to the same extent as THM results. HAA adsorptive
properties are listed in Table 4-3.
39
Table 4-3: HAA Adsorption Affinity Indicators (Speth & Miltner, 1990; World Health Organization, 2004)
Speciated THM breakthroughs along with TOC for reference are shown in
Figures 4-3 through 4-5 and the BV50 values summarized in Table 4-4. The influent THM
concentration was dominated by TCM, hence the TTHM and TCM breakthroughs trend
closely throughout all the speciated breakthrough graphs presented. The TCM
breakthrough for all three GAC types shows the chromatographic effect (normalized
concentration reaching values greater than one). Desorption due to competitive
adsorption and concentration gradient reversal has been shown to cause chromatographic
peaking in many studies (Sontheimer et al., 1988; Babi et al., 2007; Johnson et al., 2009).
The more strongly adsorbing coconut based GAC yielded the highest peak overshoot
concentration, 1.6, while the other two GACs had peak C/C0 of about 1.2. Comparison
with Table 2-3 shows that experimental breakthrough occurs in the ranges reported in
past studies.
Compound Molar Mass log Kow K 1/n q10 g/mol (μg/g)*(L/μg)^(1/n) mg/g
MCAA 94.49 0.22 0.43 0.78 0.003 MBAA 138.95 0.41 94.93 0.36 0.21 DCAA 128.94 0.92 208.83 0.30 0.42 DBAA 217.84 0.70 504.89 0.29 0.98 TCAA 168.38 1.33 704.53 0.25 1.25
40
Figure 4-3: TTHM, Speciated THM and TOC Breakthrough - Bituminous 10min EBCT
0.0
0.2
0.4
0.6
0.8
1.0
1.2
0 5,000 10,000 15,000 20,000 25,000 30,000 35,000 40,000
C/C 0
Throughput (Bed Volumes)
TTHM (Inf= 29 μg/L)
TCM (Inf= 28.2 μg/L)
DCBM (Inf = 0.85 μg/L)
TOC (Inf = 1.3 mg/L)
41
Figure 4-4: TTHM, Speciated THM and TOC Breakthrough - Lignite 10min EBCT
0.0
0.2
0.4
0.6
0.8
1.0
1.2
0 5,000 10,000 15,000 20,000 25,000 30,000 35,000 40,000 45,000
C/C 0
Throughput (Bed Volumes)
TTHM (Inf= 29 μg/L)
TCM (Inf= 28.2 μg/L)
DCBM (Inf= 0.85 μg/L)
TOC (Inf=1.3 mg/L)
42
Figure 4-5: TTHM, Speciated THM and TOC Breakthrough - Coconut 10min EBCT Table 4-4: Bed Volumes to 50% Breakthrough (BV50)
TOC TOC TTHM TCM DCBM
EBCT 5min 10min 10min 10min 10min Bituminous 12,000 14,000 15,000 15,000 NBT Lignite 11,000 12,000 18,500 18,000 NBT Coconut 2,500 3,000 21,000 21,000 NBT *NBT = No Breakthrough to 50%
As shown in Figure 4-6 and summarized in Table 4-4 the coconut-based GAC
was the best performing GAC for THM removal and the lignite-based GAC slightly
outperformed bituminous-based GAC. No DCBM breakthrough was found for the
coconut-based GAC, while 10% breakthrough occurred at about 30,000 BV for the other
43
two GACs. For both the bituminous and lignite based GACs, the BV50 values of TOC
and TCM were similar, indicating similar performance for TCM and THM precursors.
Figure 4-6: TTHM Breakthrough at 10min EBCT – Carbon Type (Inf. TTHM = 28.5 µg/L) Coconut based GACs have the highest iodine numbers, which correspond to a
higher capacity to adsorb small molecules, such as volatile organic chemicals
(Sontheimer et al., 1988). These results indicate coconut GAC to be most effective for
THM removal, but least effective for TOC removal. Due to low TOC removal, coconut
GAC is not suited for distribution system applications due to high DBP reformation
potential from early TOC breakthrough.
0.0
0.2
0.4
0.6
0.8
1.0
1.2
1.4
1.6
0 5,000 10,000 15,000 20,000 25,000 30,000 35,000 40,000
TTH
M C/C
0
Throughput (Bed Volumes)
Calgon TTHM
Lignite TTHM
Coconut TTHM
44
As shown in Figures 4-4 through 4-6 and Table 4-4, the bituminous GAC
performed the best for combined TOC and TTHM removal, and hence was chosen for
further testing the effect of source water quality.
4.2 Effect of Source Water Quality (Adsorptive RSSCT #2)
The objective of this work was to evaluate the impact of different influent
conditions in response to additional chlorination and a higher level of bromide on THM
and TOC adsorption. All three of the columns in RSSCT #2 were packed with
bituminous GAC at an EBCT of 10 min and fed three different influent waters. The three
influent waters were Boulder tap water (BT), Boulder tap water spiked up to 1 mg/L
chlorine (BTCl2), and Boulder tap spiked to 100 µg /L bromide (BTBr). Average influent
values are shown in Table 4-5.
Table 4-5: Average influent Water Characterization
BDL – below detection limit of 0.3 µg/L
The TTHMs in the BT water were 98% TCM, while the addition of chlorine
increased the TTHMs by 47% and shifted the speciation to about 80% TCM and 20%
DCBM. The addition of bromide to the BT water increased the TTHMs by 12% and
shifted the speciation to about 65% TCM, 14% DCBM, 15% DBCM and 6% TBM. Mok
et al. (2005) found that shifting of the dominant THM species from chlorinated one to
brominated one occurs at very low bromide concentration reflecting the significant
Source Water TOC TTHM TCM DCBM DBCM TBM mg/L µg/L µg/L µg/L µg/L µg/L Boulder Tap (BT) 2.2 58.5 57.5 1.1 BDL BDL
Boulder Tap + Chlorine (BTCl2) 2.3 85.9 71.4 14.6 BDL BDL Boulder Tap + Bromide (BTBr) 2.0 65.3 42.4 9.0 9.7 4.2
45
impact of bromide on THM production. Pourmoghaddas et al. (1993) reported similar
shifts in THM speciation in the presence of chlorine and bromide.
While the average values reported in Table 4-5 are good generalizations of the
influent makeup, it was found that the influent THM concentrations of all three influents
decreased over the month long experiment run time. This trend is shown in Figure 4-7,
along with a linear interpolation between points, which was used for data normalization.
Figure 4-7: Influent TTHM Concentration Gradient - Influent Chlorine and Bromide)
The impact of decreasing influent concentration on TCM breakthrough was
modeled with the PSDM and results are shown in Figures 4-8 and 4-9. Interpolated
influent values (Figure 4-7) were input to the PSDM in order to generate breakthrough
graphs, which are shown alongside experimental effluent data. TCM makes up >90% of
0 8 16 24 32
0
20
40
60
80
100
120
140
0 10,000 20,000 30,000 40,000
Experiment Run Time (Days)
TTH
M μg/L
Throughput (Bed Volumes)
BT Influent TTHM Interpolaron
BTCl2 Influent TTHM Interpolaron
BTBr Influent TTHM Interpolaron
BT Influent TTHM Data
BTCl2 Influent TTHM Data
BTBr Influent TTHM Data
46
the TTHM for the waters BT and BTCl2 and is modeled as a close representation of
TTHM.
Figure 4-8: Model and Experimental Breakthrough of TCM at 10min EBCT for the BT and BTCl2 waters
Both model and experimental data exhibit negative slopes after initial
breakthrough due to decreasing influent concentration. To facilitate more conventional
interpretation of the experimental breakthrough, the data were normalized with the linear
regression of the influent concentration values shown above in Figure 4-7. Normalizing
the data allows comparison of BV50 and BV10 values between our experimental runs and
with literature. Data normalization is shown in Figure 4-9 compared with normalized
model runs.
0
20
40
60
80
100
120
0 10,000 20,000 30,000 40,000 50,000 60,000
TCM C/C
0
Throughput (Bed Volumes)
BT TCM 10min EBCT MODEL
BT TCM 10min EBCT DATA
BTCl2 TCM 10min EBCT MODEL
BTCl2 TCM 10min EBCT DATA
47
Figure 4-9: Model and Experimental Normalized Breakthrough at 10min EBCT
Model data presented in Figure 4-9 are shown at an increased bed volume interval
for clarity. In both Figures 4-8 and 4-9, the model fits the BTCl2 data very closely but
over predicts BT breakthrough significantly. This trend is likely due to the continuing
decreasing concentration of the BT influent and the model’s inherent limitations in
predicting such a variable influent. The RSSCT #2 data henceforth presented in this
section has been normalized to the decreasing influent as shown in Figure 4-9.
4.2.1 TOC Adsorption
If GAC is to be utilized in the distribution system, then rechlorination post GAC
treatment is required as GAC reacts with the chlorine and a chlorine residual is required
at all points in the distribution system; secondary disinfection. Thus, an understanding of
the TOC breakthrough is important, as the added chlorine will react to form more DBPs.
0
0.5
1
1.5
2
2.5
3
0 10,000 20,000 30,000 40,000 50,000
TTHM
C/C
0
Throughput (Bed Volumes)
BT Model ADJUSTED
BT Data ADJUSTED
BTCl2 Model ADJUSTED
BTCl2 Data ADJUSTED
48
Bituminous GAC was used for all three columns in RSSCT #2, thus the results can be
used to assess the effects of influent bromide and chlorine concentrations on TOC
breakthrough. The TOC results at EBCTs of 5and 10 min EBCT are shown in Figures 4-
10 and 4-11, respectively.
Figure 4-10: TOC Breakthrough at 5min EBCT for three influent conditions – BT, BTCl2 and BTBr
0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1.0
0 5,000 10,000 15,000 20,000 25,000 30,000 35,000 40,000 45,000 50,000
TOC C/C 0
Throughput (Bed Volumes)
BTCl2 5min
BTBr 5min
BT 5min
49
Figure 4-11: TOC Breakthrough at 10min EBCT for three influent conditions – BT, BT with added Chlorine and BT with added Bromide Table 4-6: Bed Volumes to 50% Breakthrough (BV50) of TOC at influent TOC concentration of 2.1-2.3 mg/L
EBCT 5min 10min 20min BT 6,000 9,500 - BTCl2
6,600
9,000
-
BTBr 8,200 13,000 15,000 Zachman & Summers model
-
9,300
11,200
Bituminous RSSCT#1*
12,000
14,000
-
*Influent TOC of 1.3 mg/L
0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1.0
0 5,000 10,000 15,000 20,000 25,000 30,000 35,000 40,000 45,000
TOC C/C 0
Throughput (Bed Volumes)
BTCl2 10min BTBr 10min BT 10min
50
The effect of chlorine did not seem to be significant as the BTCl2 and BT waters
TOC breakthrough behaved similarly. One trend that persists for both EBCTs is that
bromide addition appeared to have a positive impact on TOC removal, especially in the
first half of the run. Every datum point for the BTBr water up until about 15,000 BV
shows enhanced TOC removal relative to BT and BTCl2 waters.
Figure 4-12: TOC Breakthrough at 5, 10 and 20min EBCT for the BTBr water
The TOC breakthrough at 5, 10 and 20 min are shown in Figure 4-12 for the BT
water with added bromide. The results suggest that EBCT affects TOC removal; with the
20 minute EBCT consistently showing enhanced TOC removal relative to the 5 and 10
0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1.0
0 10,000 20,000 30,000 40,000 50,000
TOC C/C 0
Throughput (Bed Volumes)
BTBr 5min
BTBr 10min
BTBr 20min
51
minute EBCTs. The TOC breakthrough results from RSSCT #1 also showed better
removal at the 10 min EBCT relative to that at 5 min.
4.2.2 DBP Removal
HAA adsorption results for adsorptive RSSCT #2, located in Appendix D, were
nonsystematic and thus not analyzed to the same extent as THM results. Low influent
HAA concentrations in both the BT and BTBr waters in addition to irregular
breakthrough in the BTCl2 water contributed to the nonsystematic nature of the data.
Speciated breakthrough for THMs are presented along with TOC for reference in
Figures 4-13 through 4-17 for each of the runs. As shown in Table 4-5, the majority of
the influent TTHM is TCM in all cases, hence the TTHM and TCM breakthroughs results
trend closely with each other. All five graphs show chromatographic effects as
normalized concentrations reach values greater than one. The 50% breakthrough values
are shown in Table 4-6. A comparison of the TTHM breakthrough for all five runs is
shown in Figure 4-18.
52
Figure 4-13: TTHM, Speciated THM and TOC Breakthrough - BT 10min EBCT
0.0
0.2
0.4
0.6
0.8
1.0
1.2
1.4
1.6
1.8
2.0
0 5,000 10,000 15,000 20,000 25,000 30,000 35,000 40,000 45,000
C/C 0
Throughput (Bed Volumes)
TCM (Inf=57.5 μg/L )
DCBM (Inf=1.1 μg/L )
TOC (Inf=2.2 mg/L )
TTHM (Inf=58.5 μg/L )
53
Figure 4-14: TTHM, Speciated THM and TOC Breakthrough - BTCl2 10min EBCT
0.0
0.2
0.4
0.6
0.8
1.0
1.2
1.4
1.6
1.8
0 10,000 20,000 30,000 40,000 50,000 60,000 70,000
C/C 0
Throughput (Bed Volumes)
TCM (Inf= 71.4 μg/L)
DCBM (Inf= 14.6 μg/L)
TOC (Inf= 2.3 mg/L)
TTHM (Inf= 85.9 μg/L)
54
Figure 4-15: TTHM, Speciated THM and TOC Breakthrough - BTBr 5min EBCT
Figure 4-16: TTHM, Speciated THM and TOC Breakthrough - BTBr 10min EBCT
0.0
0.5
1.0
1.5
2.0
2.5
0 5,000 10,000 15,000 20,000 25,000 30,000 35,000 40,000 45,000 50,000
C/C 0
Throughput (Bed Volumes)
TCM (Inf= 42.4 μg/L)
DCBM (Inf= 9 μg/L)
CDBM (Inf= 9.7 μg/L)
TOC (Inf= 2.0 mg/L)
0.0
0.2
0.4
0.6
0.8
1.0
1.2
1.4
1.6
1.8
0 10,000 20,000 30,000 40,000 50,000 60,000 70,000
C/C 0
Throughput (Bed Volumes)
TCM (Inf= 42.4 μg/L)
DCBM (Inf= 9.0 μg/L)
CDBM (Inf= 9.7 μg/L)
TBM (Inf= 4.2 μg/L)
TOC (Inf= 2.0 mg/L)
TTHM (Inf= 65.3 μg/L)
55
Figure 4-17: TTHM, Speciated THM and TOC Breakthrough - BTBr 20min
Figure 4-18: TTHM Breakthrough - Influent Chlorine and Bromide
0.0
0.2
0.4
0.6
0.8
1.0
1.2
1.4
1.6
0 5,000 10,000 15,000 20,000 25,000 30,000 35,000 40,000 45,000
C/C 0
Throughput (Bed Volumes)
TCM (Inf= 42.4 μg/L)
DCBM (Inf= 9.0 μg/L)
CDBM (Inf= 9.7 μg/L)
TOC (Inf= 2.0 mg/L)
TTHM (Inf= 65.3 μg/L)
0.0
0.2
0.4
0.6
0.8
1.0
1.2
1.4
1.6
1.8
0 10,000 20,000 30,000 40,000 50,000 60,000 70,000
C/C
0
Throughput (Bed Volumes)
BT 10min (Inf TTHM= 58.5 μg/L)
BTCl2 10min (Inf TTHM= 85.9 μg/L)
BTBr 10min (Inf TTHM= 65.3 μg/L)
56
Table 4-7: 50% Breakthrough Values (Bed Volumes x 103) TOC TTHM TCM
EBCT 5 min
10 min
20 min
5 min
10 min
20 min
5 min
10 min
20 min
Boulder Tap (BT) 6.0 9.5 -‐ -‐ 16.5 -‐ -‐ 16.5 -‐
Boulder Tap + Chlorine (BTCl2)
6.6 9.0 -‐ -‐ 6.5 -‐ -‐ 6.0 -‐
Boulder Tap + Bromide (BTBr) 8.2 13.0 15.0 14.0 16.5 14.0 11.0 14.0 14.0
DCBM CDBM TBM
EBCT 5 min
10 min
20 min
5 min
10 min
20 min
5 min
10 min
20 min
Boulder Tap (BT) -‐ 30.0 -‐ -‐ NBT -‐ -‐ NBT -‐
Boulder Tap + Chlorine (BTCl2)
-‐ NBT -‐ -‐ NBT -‐ -‐ NBT -‐
Boulder Tap + Bromide (BTBr) 29.0 30.0 25.5 NBT 31.0 NBT NBT NBT NBT
* NBT = No Breakthrough Table 4-7 is a summary of the data presented for THM and TOC removal from
RSSCT #2. Elevated influent TTHMs cause early breakthrough of TTHMs. In the
following sections, trends from the data presented above with be explored and discussed.
On average the ratio of BV50 values for TCM to TOC for the five cases above and for
RSSCT #1 with bituminous GAC, was 1.14 with a standard deviation of 0.37. This
indicates that the adsorption performance for TCM and TTHM precursors, as measured
by TOC, is similar under these conditions with prechlorination.
57
4.2.3 Effect of Influent TOC on TTHM Breakthrough
Comparison of RSSCT #1 Boulder Tap (TOC=1.3 mg/L , TTHM = 29 µg/L) and
RSSCT #2 Boulder Tap (TOC=2.2 mg/L , TTHM = 58.5 µg/L) allows the effect of
influent TOC on THM removal to be investigated as both RSSCTs were performed with
bituminous carbon. TOC and THM data from these two scenarios are graphed in Figure
4-19. No significant effect of influent TOC on THM removal is observed in Figure 4-19.
As expected, the higher TOC and THM influent water (RSSCT #2) breaks through
slightly earlier than the lower TOC and THM water (RSSCT #1). While the higher THM
water breaks through before the lower THM water, there are a few suspect points
between 15,000 and 30,000 BV where the lower THM water shows higher C/C0 values.
Since the RSSCT #2 data had to be normalized to decreasing influent concentration
values while the RSSCT #1 data were normalized to a constant influent value, some
inconsistencies such as these can be expected.
58
Figure 4-19: Effect of Influent TOC on THM Breakthrough at 10min EBCT – Boulder Tap Water from RSSCT #1 and RSSCT #2
4.2.4 Effect of EBCT on THM Breakthrough
The influent water BTBr (Boulder tap spiked with bromide) was monitored at
EBCTs of 5, 10 and 20 minutes in order to investigate the effects of EBCT on THM
breakthrough. Data from the experimental columns are shown in Figures 4-20 and 4-21.
The PSDM model was run for the same three EBCTs at the same influent concentration
and the results are shown in Figure 4-22.
0.0
0.2
0.4
0.6
0.8
1.0
1.2
1.4
1.6
1.8
0 10,000 20,000 30,000 40,000 50,000
C/C 0
Throughput (Bed Volumes)
TOC (Inf=1.3 mg/L )
TTHM (Inf=29 μg/L) TOC=1.3 mg/L TOC (Inf=2.2 mg/L )
TTHM (Inf=58.5 μg/L) TOC=2.2 mg/L
59
Figure 4-20: Experimental Effect of EBCT on TCM Breakthrough - BTBr water
Figure 4-21: Experimental Effect of EBCT on DCBM Breakthrough – BTBr water
0.0
0.2
0.4
0.6
0.8
1.0
1.2
1.4
0 5,000 10,000 15,000 20,000 25,000
C/C
0
Throughput (Bed Volumes)
BTBr 5min (Inf TCM= 42.4 μg/L)
BTBr 10min (Inf TCM= 42.4 μg/L)
BTBr 20min (Inf TCM= 42.4 μg/L)
0.0
0.2
0.4
0.6
0.8
1.0
1.2
1.4
1.6
1.8
2.0
0 5,000 10,000 15,000 20,000 25,000 30,000 35,000 40,000 45,000
C/C
0
Throughput (Bed Volumes)
BTBr 5min (Inf DCBM= 9.0 μg/L)
BTBr 10min (Inf DCBM= 9.0 μg/L)
BTBr 20min (Inf DCBM= 9.0 μg/L)
60
Figure 4-22: Single Solute Modeled EBCT Effect on TCM Breakthrough in Organic Free Water The observed effects of EBCT on TCM and DCBM breakthrough are consistent
with modeled results, showing that GAC adsorption capacity does not increase with
increasing EBCT. Results from the 5 min EBCT initially break through first but reach
total breakthrough last. The 20 min EBCT breakthrough initially breaks through last, but
has the steepest breakthrough and reaches total breakthrough first. The results from the
10 min EBCT are in between.
4.2.5 Effect of Influent Concentration on TTHM Breakthrough
The effect of influent concentration on breakthrough for microgram per liter
concentrations of THMs is presented in Figures 4-23 through 4-25 for both modeled data
and experimental data for both single-solute and co-solute scenarios.
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
8,000 10,000 12,000 14,000 16,000 18,000 20,000
C/C
0
Throughput (Bed Volumes)
TCM 42.4 μg/L 5min EBCT
TCM 42.4 μg/L 10min EBCT
TCM 42.4 μg/L 20min EBCT
61
Figure 4-23: Modeled Single-solute TCM Breakthrough at 10min EBCT at different influent concentrations The modeled single solute graph shows that higher influent concentrations of
TCM correspond to earlier breakthrough. While it is not surprising to see this trend, as it
is known to occur at mg/L concentrations. The work done by Corwin and Summers, 2012
and Summers et al., 2013 at nanogram/L concentration showed no such effect of influent
concentration.
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
0 5,000 10,000 15,000 20,000 25,000 30,000
C/C 0
Throughput (Bed Volumes)
71.4 μg TCM/L
57.5 μg TCM/L
42.4 μg TCM/L
62
Figure 4-24: Experimental TCM Breakthrough at 10min EBCT Experimental data for TCM breakthrough supports the model implications,
showing the water with the highest influent concentration of TCM (BTCl2) breaking
through first. The points at which BTBr (42.4 µg/L TCM) breaks through higher than BT
(57.5 µg/L TCM) can be explained by the higher TTHM content in the BTBr water (65.3
µg/L TTHM) relative to the BT water (58.5 µg/L TTHM). A co-solute model run
mimicking experimental influent conditions was conducted to verify this hypothesis.
0.0
0.2
0.4
0.6
0.8
1.0
1.2
1.4
1.6
1.8
0 5,000 10,000 15,000 20,000 25,000 30,000 35,000 40,000 45,000 50,000
C/C
0
Throughput (Bed Volumes)
BTCl2 10min (Inf TCM= 71.4 μg/L)
BT 10min (Inf TCM= 57.5 μg/L)
BTBr 10min (Inf TCM= 42.4 μg/L)
63
Figure 4-25: Modeled Co-Solute TCM Breakthrough at 10min EBCT at three influent concentrations The co-solute model shows that the BT and BTBr breakthrough are much closer
than in the single solute model, thus placing the conflicting experimental data within
reason.
4.2.6 Relative Effects of NOM and Co-‐solutes on THM Breakthrough
Single solute runs represent how the solute should behave if only that solute at the
specified concentration is present in the water. Experimental data and various modeling
scenarios are compared against the reference of the single solute run due to its simplicity.
Figure 4-26 shows a single solute, same concentration model output, which demonstrates
the differences in absorbability between the THMs (Table 4-2). The results in Figures 4-
13 through 4-18 and Table 4-7 indicated breakthrough of THMs according to the
expected order from the literature and Table 4-2 values, with TCM breaking through first
0
0.2
0.4
0.6
0.8
1
1.2
0 10,000 20,000 30,000 40,000 50,000
C/C 0
Throughput (Bed Volumes)
71.4 μg TCM/L
57.5 μg TCM/L
42.4 μg TCM/L
64
followed by DCBM, and DBCM. Model results, Figure 4-26, indicate that the earliest
expected breakthrough of TBM occurs at about 120,000 bed volumes, much past the
experimental RSSCT run time of 40,000 bed volumes. No breakthrough of the most
strongly adsorbing compound, TBM, occurred throughout the duration of the
experimental adsorption RSSCT.
Figure 4-26: Modeled Single-Solute THM Relative Breakthrough The next step was to graph each of the single-solute breakthrough alongside their
co-solute and NOM-solute breakthrough (Figure 4-27 and 4-28). Co-solute and NOM-
solute runs represent how the solute should behave when there is competition for
adsorption sites. NOM-solute
65
Figure 4-27: Single-solute, Co-solute Breakthrough and NOM-Solute for TCM and DBCM- PSDM Model
66
Figure 4-28: Single-solute, Co-solute Breakthrough and NOM-Solute for DCBM and TBM- PSDM Model The NOM-solute model consistently reached breakthrough first, followed by the
co-solute and single-solute model. The same trend is present for each of the THMs
modeled, indicating that the presence of both NOM and co-solutes are important to
consider when analyzing THM breakthrough.
The co-solute chromatographic effects seen in Figures 4-27 and 4-28 are similar
to experimental results and reported literature and may be a result of both competitive
adsorption and/or desorption due to concentration gradient reversal (Babi et al., 2007;
Water Research Foundation, 2009; Sontheimer et al., 1988). Desorption may occur when
67
adsorbed compounds are displaced by more strongly adsorbing compounds (competitive
adsorption), or when the concentration gradient in the adsorber reverses and adsorbed
compounds are driven into the water phase by back diffusion (Corwin & Summers,
2010). Studies by Babi et al., (2007) and Kim and Kang, (2008) report decreasing
influent TTHM values and corresponding desorption incidents due to concentration
gradient reversal. The model with NOM does not yield desorption as it is based not on
competition, but on the diminished single solute adsorption capacity.
Model and experimental TTHM values are shown alongside each other in Figures
4-29 through 4-31 and Tables 4-8 through 4-12. Experimental breakthrough trends well
with model breakthrough showing a positive relationship between model output and
experimental data. This result brings confidence to the model output with single-solute
and NOM-solute conditions tending to bound experimental breakthrough.
68
Figure 4-29: BT 10min ECBT Model and Experimental TTHM Breakthrough (Inf TTHM = 58.5 µg/L)
69
Figure 4-30: BTCl2 10min EBCT Model and Experimental Breakthrough (Inf TTHM = 85.9 µg/L)
70
Figure 4-31: BTBr 10min EBCT Model and Experimental Breakthrough (Inf TTHM = 65.3 µg/L)
Table 4-8: BT 10min EBCT Model and Experimental Breakthrough TCM (inf=57.5 µg/L) DCBM (inf= 1.1 µg/L) Modeling Condition 10% 50% 10% 50% Single-Solute BV*103 12 14 130 139 Co-Solute BV*103
12
14
31
45
Co-Solute % Diff 2.0% 0.0% 76.1% 67.6% NOM-Solute BV*103
5
5
38
42
NOM-Solute % Diff 63.1% 62.7% 70.7% 69.6% Experimental Data BV*103
6
17
24
30
Experimental Data % Diff 51.8% 25.9% 81.9% 78.4%
71
Table 4-9: BTCl2 10min EBCT Model and Experimental Breakthrough
TCM (inf= 71.4 µg/L) DCBM (inf= 14.6 µg/L) Modeling Condition 10% 50% 10% 50% Single-Solute BV*103 12 13 50 53 Co-Solute BV*103
11
12
29
37
Co-Solute % Diff 3.5% 3.2% 42.2% 30.2% NOM-Solute BV*103
4
5
16
18
NOM-Solute % Diff 61.9% 61.9% 67.6% 66.0% Experimental Data BV*103
4
6
27
-
Experimental Data % Diff 65.2% 52.0% 46.0% -
Table 4-10: BTBr 10min EBCT Model and Experimental Breakthrough TCM (inf=42.4 µg/L) DCBM (inf= 9 µg/L) CDBM (inf=9.7 µg/L) TBM (inf=4.2 µg/L) Modeling Condition 10% 50% 10% 50% 10% 50% 10% 50% Single-Solute BV*103 14 15 59 63 134 143 296 319
Co-Solute BV*103
13
14
33
42
59
63
91
108
Co-Solute % Diff 2.9% 5.4% 43.7% 34.1% 56.3% 55.9% 69.2% 66.1% NOM-Solute BV*103
5
5
19
21
42
46
86
96
NOM-Solute % Diff 63.4% 63.0% 67.8% 66.5% 68.6% 67.6% 70.9% 70.0% Exp Data BV*103
10
14
22
30
30
31
34
-
Exp Data % Diff 26.5% 5.4% 62.7% 52.4% 77.6% 78.3% 88.5% -
Table 4-11: Reverse BTBr Concentration Model Breakthrough TCM (inf= 4.2 µg/L) DCBM (inf= 9.7 µg/L) CDBM (inf=9.0 µg/L) TBM (inf=42.4 µg/L) Modeling Condition 10% 50% 10% 50% 10% 50% 10% 50% Single-Solute BV*103 30 32 57 61 138 147 136 148
Co-Solute BV*103
23
25
44
48
86
95
120
138
Co-Solute % Diff 21.3% 22.5% 23.7% 21.8% 37.4% 35.3% 12.1% 6.8% NOM-Solute BV*103
10
11
18
20
43
48
42
48
NOM-Solute % Diff 66.6% 65.6% 68.4% 67.3% 68.6% 67.7% 68.8% 67.9%
72
Table 4-12: Same Concentration Model Breakthrough TCM (inf=15 µg/L) DCBM (inf= 15 µg/L) CDBM (inf=15 µg/L) TBM (inf=15 µg/L) Modeling Condition 10% 50% 10% 50% 10% 50% 10% 50% Single-Solute BV*103 19 21 49 53 114 122 193 208
Co-Solute BV*103
17
18
37
42
84
96
131
164
Co-Solute % Diff 12.4% 13.3% 24.5% 20.0% 26.5% 21.3% 32.4% 21.1% NOM-Solute BV*103
7
7
16
18
36
40
59
65
NOM-Solute % Diff 65.5% 64.8% 67.3% 66.3% 68.2% 67.3% 69.7% 68.6%
The data presented in Figures 4-29 through 4-31 and Tables 4-8 through 4-12
allows generalization to be made about the relative effects of NOM and co-solutes on
THM breakthrough.
When THM “A” is present in significantly greater concentration than competing
THM “B” (THM “A” >> THM “B”), co-solute effects are inconsequential compared to
NOM-solute effects on the adsorption of THM “A”. In the same case, co-solute effects
must be taken into account when considering the adsorption of THM “B”. The data in
Tables 4-8 and 4-9 shows TCM at significantly greater concentrations (57.5 µg/L, 71.4
µg/L) than DCBM (1.1 µg/L, 14.6 µg/L) respectively. In these cases, the NOM-solute
model for TCM shows a much closer correlation to experimental data than the co-solute
model, while the co-solute model for DCBM shows a close correlation to experimental
data. In summary, THM “A” exerts a significant co-solute competition effect on THM
“B”, while THM “B” exerts no such effect on THM “A”. Thus, co-solute effects must be
considered when the compound of interest is present in orders of magnitude less than
other competing compounds.
Modeling all THMs at the same concentration elucidated the effect of THM
adsorbability on the co-solute and NOM-solute model outputs. Throughout both the
experimental data and modeling scenarios it is demonstrated that THM adsorbability is
73
the most important factor in determining breakthrough order, with influent concentration
determining localized breakthrough. Table 4-12 shows that for the NOM-solute and more
drastically co-solute models, the difference from the single-solute model increases with
increasing adsorbability. Weakly adsorbed compounds reach breakthrough fast and are
less affected by the breakthrough of strongly adsorbed compounds. As the strongly
adsorbed compounds reach breakthrough, they are affected by the prior breakthrough of
all the weakly adsorbed compounds. Thus, the observed co-solute effect is greater for
strongly adsorbed compounds than for weakly adsorbed compounds (Sontheimer et al.,
1988).
NOM-solute model outputs for all THMs tended to have relatively constant %
difference from the single solute outputs, generally between 60-70%. Past studies have
shown that capacity losses from preloaded carbon compared to single-solute isotherms
for many adsorbates (including TCM) are not correlated to the K and n-values of the
adsorbate (Sontheimer et al., 1988). This finding supports the modeling results, which
indicate very little correlation between THM adsorbability and effect of NOM on
breakthrough.
4.3 Effect of Temperature, Influent Bromide and Influent TOC on Biodegradation of DBPs (Pilot Runs #1 and #2) Pilot scale biofilter columns were operated over a period of two months, to
investigate the biodegradation of THMs and HAAs in aged/exhausted GAC. Three pilot
column set-ups with sample ports at 5min, 10min and 20min EBCTs were utilized in two
phases (Figure 4-32).
74
Figure 4-32: Experimental Setup In the first phase as shown in Table 4-13, the three systems were run to isolate the effect
of influent bromide (A-1, B-1, C-1) In the second phase, the system was run to isolate the
effect of influent TOC (A-2, B-2, C-2) and temperature. The same exhausted GAC was
used in the columns throughout the two pilots runs, with periodic biomass samples taken
to track microbial activity.
Table 4-13: Influent Conditions
Pilot Column Target Condition TOC pH Cl2 Resid Temp ID -‐ Run Number mg/L mg/L °C
A-‐1 0 μg/L Br 1.3 7.9* 0.54* 21 B-‐1 50 μg/L Br 1.2 7.9* 0.54* 21 C-‐1 100 μg/L Br 1.2 7.9* 0.54* 21 A-‐2 1 mg/L TOC 1.4 7.9 0.54 variable B-‐2 2 mg/L TOC 2.2 8.0 0.64 21 C-‐2 3.5 mg/L TOC 3.6 8.1 0.69 21
*Data used from Run A-‐2, unammended tap water conditions
75
Bio-GAC from the Las Vegas Valley Water District (LVVWD) was used to pack
the pilot columns, with the GAC previously being exposed to a residual of 1.5-2 mg/L
Cl2 for several years. No adsorption was observed or expected. Results presented in this
section include TOC, THM and HAA removals, which in tandem with biomass
measurements are used to produce a fit for the first-order rate equation known to apply to
the biodegradation of micropollutants (Zearley & Summers, 2012). Experiments were
also carried out to evaluate the reformation of HAAs after biotreatment, as the GAC
reacts with the residual chlorine and a chlorine residual is required at all points in the
distribution system
4.3.1 Biomass Distribution Throughout Pilot Operation
Three ATP sampling events took place throughout the duration of the pilot
testing. ATP was sampled on the first day of run 1, last day of run 1/first day of run 2
(after 1 month) and at the conclusion of run 2 (after 2 months). GAC samples were taken
at the top of the 5, 10, and 20 min EBCT columns and analyzed for ATP, which is an
indirect but well correlated measurement of biomass activity (Dowdell & Summers,
2012).
76
Figure 4-33: Biomass Distribution in Chlorinated Influent Biofilter In a biofilter treating water without a chlorine residual, expected biomass
distribution is highest at the top of the filter and lowest at the bottom of the filter where
there is not enough primary substrate to support high levels of microorganisms (Wang et
al., 1995). Chlorine is known to be toxic to microorganisms; however, Kim and Kang
(2008) showed that biodegradation could occur in GAC filters receiving prechlorinated
water, because disinfectants were reduced at the top of GAC. This trend holds true in
Figure 4-31, as the biomass concentration at the top of the 5min EBCT column where the
chlorine is being reduced is consistently lower than any point deeper (longer EBCT) in
the column where there is no longer a chlorine residual.
In filters treating substrate limited (low TOC, runs A-1, B-1, C-1, A-2) influent
waters, the biomass concentration at the top of the 10min EBCT column (average of
64,000 pg ATP/g) was consistently higher than at the top of the 20min EBCT column
0
20,000
40,000
60,000
80,000
100,000
120,000
140,000
A-‐1 Start B-‐1 Start C-‐1 Start A-‐1 End A-‐2 Start
B-‐1 End B-‐2 Start
C-‐1 End C-‐2 Start
A-‐2 End B-‐2 End C-‐2 End
Biom
ass (pg ATP
/g)
Top of 5 min EBCT Column
Top of 10 min EBCT Column
Top of 20 min EBCT Column
77
(average of 51,000 pg ATP/g), indicating that more of the biodegradable TOC uptake is
occurring in the first 10min of the column.
Examining runs A-1, B-1, C-1 and A-2 as a group shows no systematic effect of
influent bromide on biomass concentration in pilot scale columns. Runs B-2 and C-2 (2
and 3.5 mg/L TOC) show that increasing primary substrate as influent TOC produces a
significant increase in biomass concentration in pilot scale biofilters.
4.3.2 TOC Removal
Bio-GAC from the LVVWD was exhausted with respect to adsorption capacity,
as the media had been in full-scale use for several years prior to the pilot columns of this
research. Previous studies by Johnson et al., (2009), Kim and Kang (2008) and Tung et
al. (2006) demonstrate that steady state biodegradation can take from one to six months
to occur in filters containing fresh GAC. Since the influent water to the filter at LVVWD
held a residual of 1.5-2 mg/L Cl2, there was no acclimation phase needed for
biodegradation of micropollutants and TOC to begin as the microbial community was
already established in the GAC. Measuring influent and effluent TOC of the pilot system
verified this assumption. Average TOC removal (Table 4-14) for all six influent
conditions through a 20-minute EBCT is 16%, with full TOC removal results located in
Appendix E.
78
Table 4-14: TOC Removal across 20 min EBCT for all six influent conditions Pilot Column Target Condition Influent TOC 20 min EBCT
Effluent TOC 20 min EBCT Removal
ID -‐ Run Number mg/L mg/L %
A-‐1 0 μg/L Br 1.28 1.10 14 B-‐1 50 μg/L Br 1.23 1.04 15 C-‐1 100 μg/L Br 1.22 1.07 12 A-‐2 1 mg/L TOC 1.40 1.04 26 B-‐2 2 mg/L TOC 2.23 1.93 13 C-‐2 3.5 mg/L TOC 3.64 2.99 18
4.3.3 Pseudo First Order Rate Equation
Biomass growth in drinking water biofilters is sustained by uptake of primary
substrate in the form of TOC. A compound at a concentration below the threshold
concentration (Smin) needed to support primary cellular processes is defined as a
secondary substrate. In this research, influent HAAs and THMs are present at
concentrations below (Smin) and therefore are targeted for removal via co-metabolism,
which occurs when nonspecific enzymes generated by the primary substrate metabolism
biodegrade secondary substrate (Zearley & Summers, 2012).
The Michaelis-Menten relationship has been used to express the reaction rate, r,
for trace contaminant utilization in biofilters (Zearley & Summers, 2012):
• C = Contaminant Concentration [HAA] • X = Biomass Concentration [pg ATP / mL Bed] • Vmax = maximum reaction rate [ng (min·pg ATP)-1]
r = − dCdtBF
=VmaxX ⋅CKm +C#
$%
&
'(
Equation 4.1: Michaelis-Menten Reaction Kinetics
79
• Km = Michaelis-Menten constant [ng L-1] • tBF = contact time in the biofilter [min]
When the contaminant concentration is very low compared to the Michaelis constant
(C≪Km), Eqn. 4.1 can be simplified into a pseudo-first-order rate:
Equation 4.2: Pseudo First Order Rate Equation
• = Contaminant Utilization Rate Constant [mL Bed (min*pg ATP)-1]
If tBF is approximated by the EBCT and Eqn. 4.2 is integrated by tBF from 0 to EBCT and
by C from CInf to CEff results in Eqn. 4.3.
Equation 4.3: Pseudo First Order Removal Model
Integrating biomass concentration, X, over the EBCT gives the total biomass activity of
the column (Eqn. 4.4).
Equation 4.4: Total Biomass Activity
• ActivityTotal = Total Biomass Activity [(pg ATP*min)/mL Bed] Inserting Eqn. 4.4 into Eqn. 4.3 allows expression of the fraction of contaminant
remaining in the effluent as a function of the contamination utilization rate constant and
the total biomass activity of the column.
r = − dCdtBF
= k"⋅X ⋅C
k ||
CEff
C Inf
= exp(−k"⋅X ⋅EBCT )
ActivityTotal = X ⋅EBCT
80
Equation 4.5: Pseudo First Order Removal Model using Activity
4.3.4 HAA Biodegradation
HAA influent data (Table 4-15) indicates that 85% of HAAs measured
were made up of either DCAA or TCAA. This section will focus on the biodegradation of
DCAA and TCAA. Full removal results for all HAAs are shown in Appendix G.
Table 4-15: Influent HAA Concentrations
Pilot Column MCAA MBAA DCAA TCAA DBAA HAA5 % DCAA+TCAA ID -‐ Run Number μg/L μg/L μg/L μg/L μg/L μg/L
A-‐1 0.5 0.1 20.9 17.1 0.8 39.4 97% B-‐1 0.8 0.5 17.6 14.1 2.9 35.8 88% C-‐1 0.4 0.6 15.9 14.3 4.2 35.6 85% A-‐2 1.8 0.1 20.6 19.3 0.4 42.1 95% B-‐2 2.5 0.8 22.5 18.8 2.8 47.4 87% C-‐2 3.0 1.6 25.0 18.5 6.6 54.8 79%
DCAA biodegradation is shown in Figures 4-34 and 4-35 as a function of
EBCT and total biomass activity for the runs at 21°C. DCAA is known to be very
biodegradable and the results show high levels (>80%) removal consistently at a 5min
EBCT, despite a chlorinated influent (Kim & Kang 2008; Johnson et al. 2009; Zhou &
Xie, 2002; Baribeau et al., 2005). The first order model at 21°C is plotted alongside
experimental data in Figure 4-35; with increasing biomass concentration causing
increased DCAA biodegradation.
CEff
C Inf
= exp(−k"⋅ActivityTotal )
81
Figure 4-34: DCAA Removal as a function of EBCT for all six influent conditions at 21° C
0
10
20
30
40
50
60
70
80
90
100
0 5 10 15 20
Removal (%
)
Empy Bed Contact Time (min)
0 μg/L Br
50 μg/L Br
100 μg/L Br
1 mg/L TOC
2 mg/L TOC
3.5 mg/L TOC
82
Figure 4-35: DCAA Removal as a function of total biomass activity for all six influent conditions at 10 and 21 °C The DCAA removal was least for the 3.5 mg/L TOC feed, despite reporting the
highest biomass concentration and treating water with the highest influent TOC (3.7 mg
TOC /L), as shown in Figures 4-34 and 4-35. The results indicate that increased influent
primary substrate as TOC did not equate to more secondary substrate uptake. High levels
of DCAA removal observed after a 5min EBCT with marginal increases in removal
occurring with increasing EBCT/activity are what would be expected from a first order
rate of biodegradation (Zearley & Summers, 2012).
HAA data plotted as percent remaining vs. total biomass activity and fitted with
an exponential trend line produces a first order rate equation from which the
0
10
20
30
40
50
60
70
80
90
100
0 250,000 500,000 750,000 1,000,000
Removal (%
)
Total Biomass Acbvity (pg ATP*min/mL)
0 μg/L Br
50 μg/L Br
100 μg/L Br
1 mg/L TOC
1 mg/L TOC at 10° C
2 mg/L TOC
3.5 mg/L TOC
First Order Model at 21°C (k=-‐2E-‐05)
83
contamination utilization rate constant ( ) can be extrapolated for the given
environmental conditions of the experiment (Table 4-16). Figure 4-35 and 4-37 show the
experimental fit to the first order model. Data from all the pilot runs contributed to
produce the fit, as first order kinetics dictate that percent removal is only a function of
biomass concentration with no relationship to influent concentration of the contaminant
(Zearley & Summers, 2012).
Table 4-16: Extrapolated Contaminant Utilization Rate Constants
pH Temperature k'' DCAA k'' TCAA °C mL*(pg ATP*min)-‐1
8.0 21 -‐2.00E-‐05 -‐7.00E-‐06
TCAA biodegradation as a function of EBCT and total biomass activity is
shown in Figures 4-36 and 4-37. TCAA is known to be less biodegradable than DCAA
and this finding is confirmed in our results (Xie & Zhou, 2002; Baribeau et al., 2005;
Kim & Kang, 2008; Johnson et al., 2009). The first order model is plotted alongside
experimental data in Figure 4-37 with increasing biomass concentration causing
increased TCAA biodegradation.
k ||
84
Figure 4-36: TCAA Removal as a function of EBCT for all six influent conditions
0
10
20
30
40
50
60
70
80
90
100
0 5 10 15 20
Removal (%
)
Empy Bed Contact Time (min)
0 μg/L Br 50 μg/L Br 100 μg/L Br 1 mg/L TOC 2 mg/L TOC 3.5 mg/L TOC
85
Figure 4-37: TCAA Removal as a function of total biomass activity for all six influent conditions at 10 and 21 °C TCAA removal appears to conform to first order rate kinetics, with >50%
removal occurring after 5min EBCT, >80% removal after 10min and > 90% removal
after 20min EBCT. No effect of influent bromide or TOC is observed. The first order
model is only fit to the data at 21°C. Significantly slower biodegradation occurring at
10°C on an EBCT basis still appears to fit the model at 21°C due to a corresponding
decrease in biomass activity at lower temperatures. Removal data at 15 °C is not included
in Figures 4-35 and 4-37 because there is no corresponding biomass measurement for the
0
10
20
30
40
50
60
70
80
90
100
0 250,000 500,000 750,000 1,000,000
Removal (%
)
Total Biomass Acbvity (pg ATP*min/mL)
0 μg/L Br
50 μg/L Br
100 μg/L Br
1 mg/L TOC
1 mg/L TOC at 10° C
2 mg/L TOC
3.5 mg/L TOC
First Order Model at 21°C (k=-‐7E-‐06)
86
time when the biofilter was being operated at 15 °C. The effect of temperature is
expanded upon in section 4.3.5.
4.3.5 Effect of Temperature of HAA Biodegradation
Changes in temperature can significantly impact HAA biodegradation. Kim and
Kang (2008) reported an average of 99% removal of HAA5 in a GAC filter adsorber
during the warm season (April 2004–October 2004) and only 34% removal of HAA5
during the cold season (January 2005–March 2005). Significant effects of temperature on
HAA biodegradation have also been reported Wu and Xie (2005). To investigate the
effect of temperature on HAA biodegradation in our columns, a jacketed column and a
recirculating chiller controlled the temperature of column A during the second pilot run
A-2. The second pilot run lasted about 3 weeks, with no temperature control during the
first week, temperature control at 15° C during the second week and 10° C during the
third week.
87
Figure 4-38: Temperature Effects on DCAA Removal
Figure 4-39: Temperature Effects on TCAA Removal
0
10
20
30
40
50
60
70
80
90
100
0 5 10 15 20
Removal (%
)
Empy Bed Contact Time (min)
21° C 15° C 10° C
0
10
20
30
40
50
60
70
80
90
100
0 5 10 15 20
Removal (%
)
Empy Bed Contact Time (min)
21° C
15° C
10° C
88
Removal of DCAA was not significantly impacted by the decrease in temperature.
TCAA removal virtually ceased at 5min EBCT when the temperature was lowered from
21° C to 10° C. When allowed a 20min EBCT, TCAA removal still reaches about 90%
even at 10° C. The 10 C data for TCAA is shown on Figure 4-35 along with the 21 C data
and the corresponding model fit.
4.3.6 THM Biodegradation and Reformation
THM removal via biodegradation has been reported to be minimal to nonexistent
(Kim & Kang 2008, Tung et al. 2006). Our results for THM biodegradation are
nonsystematic and are reported in Appendix F.
Rechlorinated influent simulated distribution system (SDS) samples demonstrated
additional formation of THMs as would be expected. Waters with higher formation
potential (high toc, high bromide) generally showed increased formation in the influent
SDS sample relative to the instantaneous influent sample. Effluent SDS reformation
showed the same trends as the influent SDS samples. THM reformation data is located in
Appendix F.
4.3.7 HAA Reformation and Treatment Effectiveness
SDS analysis performed on influent and 20min EBCT effluent samples for each
pilot run scenario is shown below in Figures 4-4 through 4-45. The influent and influent
SDS samples are representative of what a consumer would be exposed to if no treatment
strategy were applied. The 20min EBCT effluent and 20min EBCT SDS datum are
representative of what a consumer would be exposed to immediately after biofiltration
and at the end of the distribution system after rechlorination. All six influent scenarios
89
produced similar results when rechlorinated, with DCAA and TCAA still comprising the
majority of HAA5.
HAA5 reduction and reformation results for all six scenarios are presented in
Figure 4-46. Higher reformation occurs in higher TOC influent waters, and increased
formation of DBAA is observed in waters with elevated influent TOC and bromide.
Biofiltration is an effective treatment for the reduction in HAA5 both immediately after
biofiltration as well as at the end of the distribution system, across many ranges of
chlorinated influent bromide and TOC conditions.
Figure 4-40: HAA Reformation - 0 microgram/L Br Influent
0
5
10
15
20
25
30
35
40
45
50
MCAA DCAA TCAA DBAA HAA5
HAA
Con
centrabo
n (μg/L)
Inf
20min EBCT Eff
Inf SDS
20min EBCT SDS
90
Figure 4-41: HAA Reformation - 50 microgram/L Br Influent
Figure 4-42: HAA Reformation - 100 microgram/L Br Influent
0
5
10
15
20
25
30
35
40
45
MCAA DCAA TCAA DBAA HAA5
HAA
Con
centrabo
n (μg/L)
Inf
20min EBCT Eff
Inf SDS
20min EBCT SDS
0
5
10
15
20
25
30
35
40
45
MCAA DCAA TCAA DBAA HAA5
HAA
Con
centrabo
n (μg/L)
Inf
20min EBCT Eff
Inf SDS
20min EBCT SDS
91
Figure 4-43: HAA Reformation - 1mg/L TOC Influent
Figure 4-44: HAA Reformation - 2 mg/L TOC Influen
0
5
10
15
20
25
30
35
40
45
MCAA DCAA TCAA DBAA HAA5
HAA
Con
centrabo
n (μg/L)
Inf
20min EBCT Eff
Inf SDS
20min EBCT SDS
0
10
20
30
40
50
MCAA DCAA TCAA DBAA HAA5
HAA
Con
centrabo
n (μg/L)
Inf
20min EBCT Eff
Inf SDS
20min EBCT SDS
92
Figure 4-45: HAA Reformation - 3.5 mg/L TOC Influent
Figure 4-46: HAA5 Reformation - Effect of Influent Bromide and TOC
0
10
20
30
40
50
60
MCAA DCAA TCAA DBAA HAA5
HAA
Con
centrabo
n (μg/L)
Inf
20min EBCT Eff
Inf SDS
20min EBCT SDS
0
10
20
30
40
50
60
0 μg/L Br 50 μg/L Br 100 μg/L Br 1 mg/L TOC 2 mg/L TOC 3.5 mg/L TOC
HAA
5 Co
ncen
trab
on (μ
g/L) Inf 20min EBCT Eff Inf SDS 20min EBCT SDS
93
4.4 Summary of Results
Bench scale RSSCTs and pilot scale biofilter columns were operated to evaluate
adsorptive and biological mode DBP removal in activated carbon filters. Key findings are
summarized below.
4.4.1 Adsorption
A total of six RSSCTs were carried out in order to investigate the effects of GAC
type, source water quality and EBCT on the adsorption of TOC and DBPs in treated
drinking water. Bituminous, lignite and coconut carbon packed RSSCTs were operated in
parallel, with results indicating bituminous carbon as the best performing carbon for
simultaneous TOC and DBP removal. Experimental TOC breakthrough results for the
bituminous GAC are similar to the Zachman and Summers (2010) model, which predicts
50% breakthrough at about 16,000 BV. Breakthrough of the TTHM species occurred in
order of adsorbability (TCM à DCBM à DBCM à TBM) in all RSSCTs.
Experimental HAA adsorption results were nonsystematic.
Bituminous carbon was tested further to evaluate the impact of different influent
conditions in response to additional chlorination (1mg/L Cl2) and a higher level of
bromide (100 µg /L) on TOC and DBP adsorption. The results suggest that EBCT affects
TOC removal; with the 20 min EBCT consistently showing enhanced TOC removal
relative to the 5 and 10 min EBCTs. The effect of chlorine did not seem to be significant
as the BTCl2 and BT waters TOC breakthrough behaved similarly, however, bromide
addition appeared to have a positive impact on TOC removal, especially early in the filter
run.
94
Experimental results show that adsorption with bituminous GAC is an effective
treatment strategy for the removal of TOC and TTHMs through at least 6,000 bed
volumes (42 days at 10min EBCT) and often longer depending on influent conditions.
The influent TTHMs in the Boulder tap (BT) water were 98% TCM, while the
addition of chlorine yielded more THMs and shifted the speciation to about 80% TCM
and 20% DCBM. The addition of bromide to the BT water increased the TTHMs by 14%
and shifted the speciation to about 65% TCM, 14% DCBM, 15% DBCM and 6% TBM.
The influent THM concentrations of all three influents decreased over the month long
experiment run time, so the data were normalized with the linear regression of the
influent values to allow comparison of BV50 and BV10 values between our experimental
runs and with literature.
RSSCT results were compared against results produced by the PSDM.
Experimental breakthrough trends well with PSDM model breakthrough showing a
positive relationship between model output and experimental data. Breakthrough for all
RSSCTs exhibit chromatographic effects as normalized concentrations reach values
greater than one. Chromatographic effects also appear in all co-solute model runs,
suggesting that competitive adsorption and/or desorption due to concentration gradient
reversal may be the cause. No significant effect of influent TOC on THM removal is
observed between runs performed at 1.3 and 2.2 mg/L TOC. The observed effects of
EBCT on TCM and DCBM breakthrough are consistent with modeled results, showing
that GAC adsorption capacity on per bed volume basis does not increase with increasing
EBCT. All experimental and model scenarios demonstrate that THM adsorbability is the
most important factor in determining breakthrough order, with influent concentration
95
determining localized breakthrough. Modeled and experimental results indicate a
significant effect of influent concentration on breakthrough of TTHMs in the
microgram/L range. Elevated influent TTHMs produced faster breakthrough of TTHMs.
Modeled single-solute and NOM-solute conditions tend to bound experimental
breakthrough for the three RSSCTs modeled with the PSDM. The NOM-solute model
consistently reached breakthrough first, followed by the co-solute and single-solute
model. The same trend is present for each of the THMs modeled, indicating that the
presence of both NOM and co-solutes are important to consider when analyzing THM
breakthrough. For the NOM-solute and more drastically co-solute models, the difference
from the single-solute model increases with increasing adsorbability. Thus, the observed
co-solute effect is greater for strongly adsorbed compounds than for weakly adsorbed
compounds. Model results show that co-solute effects must also be considered when the
compound of interest is present in orders of magnitude less than other competing
compounds.
4.4.2 Biodegradation
Three experimental pilot scale biofiltration setups were operated under a total of
six different influent conditions. Columns were packed with exhausted bio-GAC that was
acclimated to influent chlorine residual. An average TOC removal of 16% occurred
across all six influent scenarios. THM biodegradation results were nonsystematic. DCAA
and TCAA made up >85% of HAA5 and therefore DCAA and TCAA biodegradation
were investigated further. Biodegradation of HAAs in pilot scale columns followed
expected trends from the first order model shown to apply to biodegradation of
micropollutants by Zearley and Summers (2012). Experimental DCAA removal between
96
83%-97% was reported at all EBCTS (5, 10 and 20min) for the duration of the pilot run.
TCAA removal ranged between 50%-78% at 5 minute EBCT, 80%-96% at 10 minute
EBCT and 93%-98% at 20 minute EBCT. No observed effect of influent TOC or
bromide on removal of HAAs reported. Higher temperature produced faster
biodegradation of TCAA and lower temperature significantly slowed biodegradation of
TCAA, although 90% removal was still achieved at a 20min EBCT.
HAA reduction and reformation data for all six scenarios indicated that
biofiltration is an effective treatment for the reduction in HAA5 both immediately after
biofiltration as well as at the end of the distribution system, across many ranges of
chlorinated influent bromide and TOC conditions.
97
Chapter 5 Summary and Recommendations
The goal of this project was to develop and evaluate the use of GAC in the
distribution system to meet DBPs (especially HAAs) regulations under both adsorptive
and biological modes. It was hypothesized that a post-treatment reactor strategically
located in the distribution system will offer small systems a cost-effective alternative to
controlling THMs, HAA5s and other unregulated DBPs. To verify our hypothesis, a total
of six adsorptive bench scale RSSCTs and three pilot scale biofilters were operated in
order to investigate the effects of GAC type, source water quality and EBCT on the
adsorption and biodegradation of TOC and DBPs in treated drinking water.
Bituminous, lignite and coconut carbon packed RSSCTs were operated in parallel,
with results indicating bituminous carbon as the best performing carbon for simultaneous
TOC and DBP removal. Experimental TOC breakthrough results for the bituminous GAC
are similar to the Zachman and Summers (2010) model, which predicts 50%
breakthrough at about 16,000 BV. Breakthrough of the TTHM species occurred in order
of adsorbability (TCM à DCBM à DBCM à TBM) in all RSSCTs. Experimental
HAA adsorption results were nonsystematic.
Bituminous carbon was tested further to evaluate the impact of different influent
conditions in response to additional chlorination (1mg/L Cl2) and a higher level of
bromide (100 µg /L) on TOC and DBP adsorption. The results suggest that EBCT affects
TOC removal; with the 20 minute EBCT consistently showing enhanced TOC removal
relative to the 5 and 10 minute EBCTs. The effect of chlorine did not seem to be
significant as the BTCl2 and BT waters TOC breakthrough behaved similarly, however,
98
bromide addition appeared to have a positive impact on TOC removal, especially early in
the filter run.
Experimental results show that adsorption with bituminous GAC is an effective
treatment strategy for the removal of TOC and TTHMs through at least 6,000 bed
volumes (42 days at 10min EBCT) and often longer depending on influent conditions.
RSSCT results were compared against results produced by the PSDM.
Experimental breakthrough trends well with PSDM model breakthrough showing a
positive relationship between model output and experimental data. Breakthrough for all
RSSCTs exhibit chromatographic effects as normalized concentrations reach values
greater than one. Chromatographic effects also appear in all co-solute model runs,
suggesting that competitive adsorption and/or desorption due to concentration gradient
reversal may be the cause. No significant effect of influent TOC on THM removal is
observed between runs performed at 1.3 and 2.2 mg/L TOC. The observed effects of
EBCT on TCM and DCBM breakthrough are consistent with modeled results, showing
that GAC adsorption capacity on per bed volume basis does not increase with increasing
EBCT. All experimental and model scenarios demonstrate that THM adsorbability is the
most important factor in determining breakthrough order, with influent concentration
determining localized breakthrough. Modeled and experimental results indicate a
significant effect of influent concentration on breakthrough of TTHMs in the
microgram/L range. Elevated influent TTHMs produced faster breakthrough of TTHMs.
Modeled single-solute and NOM-solute conditions tend to bound experimental
breakthrough for the three RSSCTs modeled with the PSDM. The NOM-solute model
consistently reached breakthrough first, followed by the co-solute and single-solute
99
model. The same trend is present for each of the THMs modeled, indicating that the
presence of both NOM and co-solutes are important to consider when analyzing THM
breakthrough. For the NOM-solute and more drastically co-solute models, the difference
from the single-solute model increases with increasing adsorbability. Thus, the observed
co-solute effect is greater for strongly adsorbed compounds than for weakly adsorbed
compounds. Model results show that co-solute effects must also be considered when the
compound of interest is present in orders of magnitude less than other competing
compounds.
Operational recommendations for adsorptive THM removal include lead-lag
operation with TOC monitoring, split stream treatment, and determination of influent
THM speciation. GAC should be installed in a lead-lag configuration (two GAC
contactors in series) for adsorptive removal of THMs. Monitoring TOC breakthrough as a
surrogate for THM breakthrough at a sample point located after the primary contactor
and prior to the secondary contactor is a cost effective way to determine when the
primary contactor GAC needs replacement, while maintaining treatment redundancy in
the secondary contactor. In such an arrangement, high levels (>90%) of THM removal
would be expected, with chromatographic peaking abated by the redundancy in
treatment. Such high levels of treatment are usually excessive to meet the stage 2 DBPR
MCLs. In order to extend GAC life while meeting regulatory limits, each water system
should determine an appropriate design flow to split off from the main distribution
system to treat in the GAC contactor system. The amount of flow treated should account
for variability in distribution flow, with regulatory limits being met at high flows and
enhanced treatment provided during lower flows. Analysis of site-specific influent THM
100
speciation should also be conducted at all potential implementation sites. Experimental
and modeled results indicate that brominated THM species are removed far more
effectively than chloroform via GAC adsorption. In treated water with high levels of
chloroform, air stripping might be a better choice due to the high volatility of lower
molecular weight THMs.
Three experimental pilot scale biofiltration setups were operated under a total of
six different influent conditions. Columns were packed with exhausted bio-GAC that was
acclimated to influent chlorine residual. An average TOC removal of 16% occurred
across all six influent scenarios. THM biodegradation results were nonsystematic. DCAA
and TCAA made up >85% of HAA5 and therefore DCAA and TCAA biodegradation
were investigated further. Biodegradation of HAAs in pilot scale columns followed
expected trends from the first order model shown to apply to biodegradation of
micropollutants by Zearley and Summers (2012). Experimental DCAA removal between
83%-97% was reported at all EBCTS (5, 10 and 20min) for the duration of the pilot run.
TCAA removal ranged between 50%-78% at 5 minute EBCT, 80%-96% at 10 minute
EBCT and 93%-98% at 20 minute EBCT. No observed effect of influent TOC or
bromide on removal of HAAs reported. Higher temperature produced faster
biodegradation of TCAA and lower temperature significantly slowed biodegradation of
TCAA, although 90% removal was still achieved at a 20min EBCT.
HAA reduction and reformation data for all six scenarios indicated that
biofiltration is an effective treatment for the reduction in HAA5 both immediately after
biofiltration as well as at the end of the distribution system, across many ranges of
chlorinated influent bromide and TOC conditions.
101
Future research on adsorptive and biological mode DBP removal in activated
carbon filters should include pilot scale operation and monitoring at critical points in a
distribution system that is currently out of compliance. The choice of operation in
adsorptive mode versus biodegradation mode should be dependent on system specific
compliance needs. Cost analysis with consideration of carbon density is recommended
for systems considering GAC for THM removal. Referring Table 3-1, lignite coal (0.39
g/cm3) is significantly less dense than both bituminous coal (0.54 g/cm3) and coconut
shell (0.50 g/cm3) GACs. The results of adsorptive RSSCT #1 show similar performance
to the bituminous GAC for TOC and THM removal on an EBCT basis, indicating that
lignite GAC could potentially provide similar treatment at a cost lower than of
bituminous GAC, as GAC is sold by weight. Additionally, a biomass acclimation study in
GAC filters under chlorinated conditions would be a significant contribution to the
literature. Important variables in this proposed study include influent temperature, TOC
and chlorine concentration. The research presented in this thesis indicates that a post-
treatment reactor strategically located in the distribution system will offer small systems a
cost-effective alternative to controlling THMs, HAA5s and other unregulated DBPs.
102
Works Cited Babi, K. G., Koumenides, K. M., Nikolaou, A. D., Makri, C. A., Tzoumerkas, F. K., & Lekkas, T. D. (2007). Pilot Study of the Removal of THMs, HAAs and DOC from Drinking Water by GAC Adsorption. Desalination , 210, 215-224. Baribeau, H., Krasner, S. W., Chinn, R., & Singer, P. C. (2005). Impact of biomass on the stability of HAAs and THMs in a simulated distribution system. Journal American Water Works Association, 97 (2), 69-81. Bayless, W., & Andrews, R. (2008). Biodegradation of Six Haloacetic Acids in Drinking Water. Journal of Water and Health , 6 (1), 15-22. Chuang, Y.-H., Wang, G.-S., & Tung, H.-H. (2011). Chlorine Residuals and Haloacetic Acid Reduction in Rapid Sand Filtration. Chemosphere , 85 (7), 1146-1153. Corwin, C. J., & Summers, R. S. (2012). Controlling trace organic contaminants with GAC adsorption. Journal American Water Works Association (104), 36-47. Corwin, C. J., (2010). Trace Organic Contaminant Removal from Drinking Waters by Granular Activated Carbon Adsorption, Desorption, and the Effect of Background Organic Matter. PhD dissertation, University of Colorado, CEAE Department, Boulder. Crittenden, J. C., Berrigan, J. K., & Hand, D. W. (1986 Design of rapid small-scale adsorption tests for a constant diffusivity. Journal Water Pollution Control Federation , 312–319. Crittenden, J. C., Trussell, R. R., Hand, D. W., Kerry , H. J., & Tchobanoglous, G. (2012). Water Treatment: Principles and Design (Third ed.). Hoboken, New Jersey: John Wiley & Sons, Inc. DeMarco, J., & Brodtmann, N. (1984). Prediction of Full Scale Plant Performance from Pilot Columns. In N. C. Society, P. V. Roberts, R. S. Summers, S. Regli, R. Pickford, & F. Bell (Eds.), Adsorption Techniques in Drinking Water Treatment (pp. 295-328). Reston, Virginia, USA: USEPA. DiGiano, F. (1983). Adsorption of Organic Substances in Drinking Water. Dowdell, K., (2012). Trace Organic Contaminant Removal in Drinking Water Biofilters under Carbonaceous and Nitrogen-Supplemented Conditions and Evaluating Biomass with ATP and Phospholipid Methods. Masters Thesis, University of Colorado, CEAE Department, Boulder.
103
Fokken, B., & Kurtz, R. (1984). Removal of purgeable organic chlorine compounds by activated carbon adsorption. In P. V. Roberts, R. S. Summers, & S. Regli, In: Adsorption Techniques in Drinking Water Treatment (EPA 570/9-84-005 ed.). Washington, D.C.: US Environ. Protection Agency. International Agency for Research on Cancer. (2014, July). IARC Monographs. Retrieved July 30, 2014, from International Agency for Research on Cancer: www.monographs.iarc.fr Johnson, B. A., Lin, J. C., Rexing , D., Fang, M., Chan , J., Jacobsen, L., et al. (2009). Localized treatment for disinfection by-products. Denver: Water Research Foundation. Kempisty, D. M. (2014, July). Adsorption of Volatile and Perfluorinated Compounds from Groundwaters using Granular Activated Carbon. PhD dissertation, University of Colorado, CEAE Department, Boulder. Kim, J., & Kang, B. (2008). DBPs Removal in GAC Filter-adsorber. Water Research, 42.1 (2), 145-152. Liu, W., & Andrews, S. A. (2001). Full-scale Adsorption of HAA5 by Activated Carbon. WQTC. Nashville, Tenn.: American Water Works Association. McCarty, P. L., Argo, D., & Reinhard, M. (1987). Operational Experiences with Activated Carbon Adsorbers at Water Factory 21. Journal Environmental Pathology, Toxicology and Oncology. , 7, 319-338. McGuire, M., Marshall, D., Tate, C., Aieta, & Ho, E. (1991). Evaluating GAC for Trihalomethane Control. Journal American Water Works Association , 83 (1), 38- 48. Meijers, A.P. et al. (1984). Objectives and Procedures for GAC Treatment in the Netherlands. In N. C. Society, P. V. Roberts, R. S. Summers, S. Regli, R. Pickford, & F. Bell (Eds.), Adsorption Techniques in Drinking Water Treatment (pp. 137-167). Reston, Virginia, USA: USEPA. Miller, R. (1984). Treatment of Ohio River Water. In N. C. Society, P. V. Roberts, R. S. Summers, S. Regli, R. Pickford, & F. Bell (Eds.), Adsorption Techniques in Drinking Water Treatment (pp. 374-394). Reston, Virginia, USA: USEPA. Mok, K. M., Wong, H., & Fan, X. J. Modeling Bromide Effects on the Speciation of Trihalomethans Formation. Global NEST Journal (7), 1-16.
104
NATO Committe on the Challenges of Modern Society. (1984). Adsorption Techniques in Drinking Water Treatment. In P. V. Roberts, R. S. Summers, S. Regli, R. Pickford, & F. Bell (Ed.). USEPA. Noto, A. (2016, May 4). Adsorption Capacity. Retrieved August 28, 2016, from Alberto Noto Recycling : http://www.notorecycling.us/removal/adsorption-capacity.html Palmdale Water District. (2011). Palmdale, CA Water District Chooses GAC Treatment to Meet TTHM Guidelines for Today and Tommorrow. Palmdale: Calgon Carbon Corporation. Potwara, R. (2012, March). The ABCs of Activated Carbon. Water Quality Products , 14,15. Pourmoghaddas, H. (1993). Effect of bromide ion on formation of HAAs during chlorination. Journal American Water Works Association , 85 (1), 82-87. Roberts, P. V., & Summers, R. S. (1982). Granular activated carbon performance for organic carbon removal. Journal American Water Works Association, 74 (2), 113- 118. Singer, P. (1994). Control of Disinfection By-Products in Drinking Water. Journal of Environmental Engineering , 120, 727-744. Sontheimer, H., Crittenden, J. C., & Summers, R. S. (1988). Activated Carbon for Water Treatment. Karlsruhe, Germany: DVGW-Forschungsstelle, Engler-Bunte-Institut, Universitat Karlsruhe (TH). Speth, T. F., & Miltner, R. J. (1990). Technical Note: Adsorption Capacity of GAC for Synthetic Organics. Journal American Water Works Association , 82 (2), 72-75. Speth, T., & Miltner, R. (1998). Adsorption Capacity of GAC for Synthetic Organics. Journal American Water Works Association , 90 (4), 171. Summers, R. S., Kim, S. M., Shimabuku, K., Chae, S. H., & Corwin, C. J. (2013). Granular activated carbon adsorption of MIB in the presence of dissolved organic matter. Water Research , 47 (10), 3507-3513. Summers, R. S., Knappe, D., & Snoeyink, V. L. (2010). Adsorption of Organic Compounds by Activated Carbon. In A. W. Association, Water Quality and Treatment (Sixth Edition ed.). New York: McGraw-Hill. Tung, H.-H., Unz, R., & Xie, Y. F. (2006). HAA removal by GAC adsorption. Journal American Water Works Association , 98 (6), 107-112.
105
USEPA. (2015, August 28). Stage 1 and Stage 2 Disinfectants and Disinfection Byproducts Rules. Retrieved 2016, from US Environmental Protection Agency : https://www.epa.gov/dwreginfo/stage-1-and-stage-2-disinfectants-and- disinfection-byproducts-rules USEPA. (1996). ICR Manual for Bench- and Pilot-Scale Treatment Studies. 814-B-96- 003 . Cincinnati, OH: EPA. Wang, J. Z., Summers, R. S., & Miltner, R. J. (1995). Biofiltration Performance: Part 1. Relationship to Biomass. Journal American Water Works Association , 87 (12), 55-63. Wood, P. R., & DeMarco, J. (1984). Treatment of Groundwater with Granular Activated Carbon. In N. C. Society, P. V. Roberts, R. S. Summers, S. Regli, R. Pickford, & F. Bell (Eds.), Adsorption Techniques in Drinking Water Treatment (pp. 348- 373). Reston, Virginia, USA: USEPA. World Health Organization. (2004). Trihalomethanes in Drinking-water: Background Document for Development of WHO Guidelines for Drinking-water Quality. Wu, H., & Xie, Y. F. (2005). Effects of EBCT and Water Temperature on HAA Removal Using BAC. Journal American Water Works Association , 97 (11), 94-101. Wu, M. (2012). Disinfectants and Disinfection Byproducts Rule (Stage 1&2 DBPRs). Wyoming Potable Water Age, Lagoon Aeration and Utility Line Replacement Seminar. EPA Region 8. Xie, Y. F., & Zhou, H. (2002). Use of BAC for HAA removal - Part 2, column study. Journal American Water Works Association , 94 (5), 126-134. Zachman, B. A., & Summers, R. S. (2010). Modeling TOC Breakthrough in Granular Activated Carbon Adsorbers. Journal of Environmental Engineering , 136 (2), 204-210. Zearley, T. L., & Summers, R. S. (2012). Removal of trace organic micropollutants by drinking water biological filters. Environmental Science and Technology , 46 (17), 9412-9419. Zearley, T., (2012). Removal of Trace Organic Micropollutants by Drinking Water Biological Filters. University of Colorado, PhD dissertation CEAE Department, Boulder.
106
Appendix A – GAC Manufacturer Specifications
107
108
Safety Message
Wet activated carbon can deplete oxygen from air in enclosed spaces. If use in an enclosed space is required, procedures for work in an oxygen deficient environment should be followed.
1.800.4CARBON calgoncarbon.com
© Copyright 2015 Calgon Carbon Corporation, All Rights ReservedDS-FILTRA40015-EIN-E1
Design Considerations
FILTRASORB 400 activated carbon is typically applied in down-flow packed-bed operations using either pressure or gravity systems.Design considerations for a treatment system is based on the user’s operating conditions, the treatment objectives desired, and the chemical nature of the compound(s) being adsorbed.
Typical Pressure DropBased on a backwashed and segregated bed
Typical Bed Expansion During BackwashBased on a backwashed and segregated bed
109
���������������2#0�o���<BBB:=��3*7�:B9>
���������
�� -2��-0'2��!2'4�2#"���0 -,�'1��.0#+'#0��!2'4�2#"�!�0 -,�+�,3$�!230#00#1.#!2#"�$-0�#6.#0'#,!#"�.#-.*#I"'4#01#�.0-"3!21��,"�120-,%�!312-+#00#*�2'-,1&'.1G��� -2U1�&'12-07�-$',,-4�2'-,I�.0-"3!2�.#0$-0+�,!#I2#!&,'!�*�#6.#02'1#��,"�!312-+#0�$-!31#,130#�2&�2�7-3�0#!#'4#�2&#�0'%&2.0-"3!21��,"�1-*32'-,1�$-0�7-30�1.#!'$'!.30'$'!�2'-,�,##"1G
����������f�<BBB �0�,3*�0���!2'4�2#"����0 -,�
�
�����������<BBB�'1��,��!'"�5�1&#"�!�0 -,�"#1'%,#"�$-0�5�2#0�20#�2+#,2�..*'!�2'-,1G� 2�'1�.0-"3!#"� 7�&'%&�2#+.#0�230#�12#�+��!2'4�2'-,�-$�*'%,'2#�!-�*G�����������<BBB�&�1���5'"#�.-0#�1'8#�"'120' 32'-,��,"�*�0%#�.-0#�4-*3+#G��#!&�0�!2#0'12'!1�.0-4'"#������������<BBB�5'2&�0�.'"��"1-0.2'-,�0�2#��,"�&'%&!�.�!'27�$-0�"'11-*4#"�-0%�,'!1G
�����������<BBB�'1��-1��!#02'$'#"I�+##21���o��� ��2�,"�0"�>9��,"�2�#/3'0#+#,21�-$�2&#�*�2#12�4#01'-,�-$�2&#��G�G�--"��&#+'!�*1��-"#6G
�-2� *#���2#0�����������<BBB��"1-0 1�2�12#I�-"-0I�!-*-0��,"�2-6'!�-0%�,'!�!-+.-3,"1�$0-+"0',)',%�5�2#0G� 2�&�1�2&#�&'%�!�.�!'27�-$��,7�!-++#0!'�*�5�2#0�!�0 -,�$-0�2�,,'!�,"�&3+'!�!-+.-3,"1I�5&'!&��0#�.0#!301-01�$-0�20'&�*-+#2&�,#�s���t�$-0+�2'-,G�����������<BBB�#$$#!2'4#*7�0#+-4#1�.#12'!'"#1I�� '!'"#1I�17,2'!�-0%�,'!!&#+'!�*1I��,"�-2��131.#!2#"�!�0!',-%#,1�$-3,"�',�"0',)',%�5�2#0�13..*'#1G�����������<BBB�+##21��**�������>B<�12�,"�0"1�$-0��!2'4�2#"�!�0 -,�$-0�0�.'"%0�4'27�$'*2#01��,"�.0#1130#�!-,2�!2-01�31#"�',�.-2� *#�5�2#0�.30'$'!�2'-,�1712#+1G
��12#5�2#0��,"��-,2�+',�2#"��0-3,"5�2#0����..*'#"��2�.-',2�1-30!#�-0�',���.-*'1&',%�$'*2#0I������������<BBB�!�,� #31#"�2-�+##2�"'1!&�0%#�*'+'21�$-0�+-12�0#%3*�2#"�-0%�,'!1G��&#� 0-�"�.-0#�1'8#"'120' 32'-,�-$������������<BBB��**-51�20#�2+#,2�-$�!-+.*#6�5�12#5�2#0�120#�+1!-,2�',',%� -2&�*�0%#��,"�1+�**�+-*#!3*�0�1'8#�.-**32�,21G��&#�+�!0-.-0-31�1203!230#-$������������<BBB�.0-4'"#1�&'%&�2-*#0�,!#�$-0�,�230�*�-0%�,'!�+�22#0�s���tI�5&'!&',2#0$#0#1�5'2&��"1-0.2'-,�-$�-2��-0%�,'!�!-,2�+',�,21G
��� ��� ���
��������������� -"',#�,3+ #0 +',G�=BB L
����������������-*�11#1�"#!-*-0'8',%�#$$'!'#,!7�s�t +',G�AB L
���������������� 0�1'-,�0#1'12�,!#�s����t +',G�?B �o++
�����������������02'!*#�1'8#�z�9B�+#1&�s:GBB�++t +�6G�= �
�����������������02'!*#�1'8#�y;B�+#1&�sBG>B�++t +�6G�= �
���������������$$#!2'4#�1'8# +',G�BG> ++
+�6G�BG@ ++
����������������,'$-0+'27�!-#$$'!'#,2 +�6G�:G9 L
����������������-'1230#�s�1�.�!)#"t +�6G�@ �
����������������312 +�6G�BG;B �
110
111
112
113
114
115
Appendix B – TOC Adsorption
116
117
118
119
120
121
122
Appendix C – THM Adsorption
123
124
125
126
Appendix D – HAA Adsorption
127
128
129
Appendix E – TOC Biodegradation
130
131
Appendix F – THM Biodegradation and Reformation
132
133
Appendix G – HAA Biodegradation and Reformation
134
135
136
137
138
139
140
Appendix H – ATP Biomass Measurements and Method
141
142